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,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,1,balanced,0.0609386662642161
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,1,balanced,0.08664533495903015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,1,balanced,0.1397119959195455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,1,balanced,0.1405280033747355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,1,balanced,0.14357333381970724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,1,balanced,0.14448533455530801
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,1,balanced,0.18014933665593466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,1,balanced,0.1814240018526713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,1,balanced,0.18037333091100058
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,1,balanced,0.18344533443450928
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,1,balanced,0.18491733074188232
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,1,balanced,0.1951733430226644
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,1,balanced,0.1962826649347941
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,1,balanced,0.20115200678507486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,1,balanced,0.3660320043563843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,1,balanced,0.2977866729100545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,1,balanced,0.22822932402292886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,1,balanced,0.5953386624654134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,1,power_law_1.01,0.05754240155220032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,1,balanced,0.2712799906730652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,1,balanced,0.36979734897613525
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,1,power_law_1.01,0.07462400197982788
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,1,balanced,0.45848000049591064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,1,power_law_1.01,0.09088000059127807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,1,balanced,0.659173329671224
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,1,power_law_1.01,0.13874560594558716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,1,power_law_1.01,0.14417920112609864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,1,balanced,0.8525760173797607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,1,power_law_1.01,0.15839359760284424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,1,balanced,1.2503680388132732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,1,balanced,0.14154133200645447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,1,balanced,0.10291733344395955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,1,balanced,1.6493493715922039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,1,balanced,2.552720069885254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,1,power_law_1.01,0.13875839710235596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,1,power_law_1.01,0.1003648042678833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,1,power_law_1.01,0.18288639783859253
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,1,balanced,3.516640027364095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,1,power_law_1.01,0.1999295949935913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,1,power_law_1.01,0.1775936007499695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,1,power_law_1.01,0.1847488045692444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,1,power_law_1.01,0.20833280086517333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,1,power_law_1.01,0.19438719749450684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,1,power_law_1.01,0.20371201038360595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,1,power_law_1.01,0.22621440887451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,1,power_law_1.01,0.22675199508666993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,1,power_law_1.01,0.23556480407714844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,1,power_law_1.01,0.25985920429229736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,1,power_law_1.01,0.30329599380493166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,1,power_law_1.01,0.33118720054626466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,1,power_law_1.01,0.4465343952178955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,1,balanced,0.14891733725865683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,1,balanced,0.2531893253326416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,1,power_law_1.2,0.14239360094070436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,1,power_law_1.2,0.21194241046905518
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,1,power_law_1.01,0.5630591869354248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,1,power_law_1.2,0.27159039974212645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,1,power_law_1.01,0.1431872010231018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,1,power_law_1.01,0.2264319896697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,1,power_law_1.01,0.17256319522857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,1,power_law_1.01,0.2764607906341553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,1,balanced,0.23884799083073935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,1,balanced,0.47650667031606037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,1,power_law_1.01,0.7524223804473877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,1,balanced,0.25643734137217206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,1,balanced,0.2560693422953288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,1,balanced,0.2686506708463033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,1,power_law_1.2,0.47747201919555665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,1,power_law_1.01,0.24677760601043702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,1,power_law_1.2,0.47812480926513673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,1,power_law_1.01,0.2552191972732544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,1,power_law_1.01,0.4656576156616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,1,power_law_1.2,0.5128384113311768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,1,power_law_1.01,0.27877120971679686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,1,power_law_1.01,0.46662402153015137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,1,balanced,0.4793119827906291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,1,balanced,0.47993067900339764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,1,balanced,0.48599998156229657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,1,power_law_1.01,0.9515775680541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,1,power_law_1.01,0.5024831771850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,1,balanced,0.3072693347930908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,1,balanced,0.3303520083427429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,1,power_law_1.2,0.4791552066802979
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,1,power_law_1.01,1.3379008293151855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,1,power_law_1.2,0.506873607635498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,1,power_law_1.01,0.3368256092071533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,1,power_law_1.01,0.3507776021957397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,1,balanced,0.4658133188883464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,1,balanced,0.46568532784779865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,1,power_law_1.01,0.46802558898925783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,1,power_law_1.01,0.4794943809509277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,1,power_law_1.01,1.7449151992797851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,1,power_law_1.2,0.5591360092163086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,1,balanced,0.30195732911427814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,1,power_law_1.2,0.594214391708374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,1,balanced,0.5342986583709717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,1,balanced,0.5447839895884196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,1,balanced,0.33452800909678143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,1,power_law_1.01,0.3716991901397705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,1,power_law_1.01,0.5900032043457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,1,power_law_1.01,0.5766719818115235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,1,power_law_1.01,0.3739327907562256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,1,power_law_1.2,0.5348671913146973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,1,balanced,0.5070186853408813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,1,power_law_1.01,0.5352000236511231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,1,balanced,0.28235199054082233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,1,power_law_1.01,0.2918207883834839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,1,power_law_1.2,0.570854377746582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,1,power_law_1.2,0.5687679767608642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,1,power_law_1.2,0.6714560031890869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,1,balanced,0.5131946802139282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,1,power_law_1.2,0.7831871986389161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,1,balanced,0.5182666778564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,1,balanced,0.5321600039800009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,1,balanced,0.7411573727925619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,1,power_law_1.2,0.8513152122497558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,1,balanced,0.7547573248545328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,1,balanced,0.7627573013305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,1,power_law_1.2,0.9323967933654785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,1,power_law_1.01,0.5812479972839355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,1,power_law_1.01,0.5890944004058838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,1,power_law_1.01,2.501036834716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,1,balanced,0.2857919931411743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,1,balanced,0.28961066404978436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,1,balanced,0.3012160062789917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,1,balanced,0.41784000396728516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,1,balanced,0.42396799723307294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,1,balanced,0.43091734250386554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,1,power_law_1.01,0.3069888114929199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,1,power_law_1.01,0.3280384063720703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,1,power_law_1.01,0.36014719009399415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,1,power_law_1.01,0.7030208110809326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,1,power_law_1.01,0.4402624130249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,1,power_law_1.01,0.7638207912445069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,1,power_law_1.01,0.47139201164245603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,1,power_law_1.01,0.530079984664917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,1,power_law_1.01,0.8593343734741211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,1,power_law_1.01,0.9617024421691894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,1,balanced,1.016586701075236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,1,balanced,1.0568640232086182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,1,power_law_1.2,1.0328191757202148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,1,power_law_1.01,1.065567970275879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,1,balanced,0.1360373298327128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,1,balanced,0.5714613199234009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,1,balanced,0.6035146713256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,1,power_law_1.01,1.3352191925048829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,1,power_law_1.01,0.13327360153198242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,1,power_law_1.2,1.3201279640197754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,1,power_law_1.01,0.5959807872772217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,1,power_law_1.01,0.7655807971954346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,1,power_law_1.01,3.458099365234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,1,balanced,1.3727466265360515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,1,power_law_1.01,1.5437888145446776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,1,balanced,1.7197759946187336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,1,power_law_1.2,1.513862419128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,1,balanced,2.539189338684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,1,balanced,3.347813288370768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,1,power_law_1.01,1.9574207305908202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,1,balanced,0.7850346565246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,1,balanced,0.9874026775360107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,1,balanced,0.1872426668802897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,1,balanced,5.15282662709554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,1,balanced,0.2849493424097697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,1,balanced,1.443461259206136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,1,power_law_1.01,0.902950382232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,1,balanced,1.8722453117370605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,1,power_law_1.2,0.10182399749755859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,1,power_law_1.2,0.1404736042022705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,1,power_law_1.2,1.8829120635986327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,1,power_law_1.2,0.16269439458847046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,1,power_law_1.2,0.2510591983795166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,1,power_law_1.2,0.25633280277252196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,1,power_law_1.01,0.17902719974517822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,1,power_law_1.2,0.27521278858184817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,1,power_law_1.01,0.20074241161346434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,1,power_law_1.2,0.3369983911514282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,1,power_law_1.2,0.3490943908691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,1,power_law_1.2,0.35727999210357664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,1,balanced,2.8864692052205405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,1,power_law_1.2,0.38463358879089354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,1,balanced,6.945503870646159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,1,power_law_1.2,0.29517440795898436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,1,power_law_1.01,1.0919487953186036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,1,power_law_1.2,0.3108479976654053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,1,balanced,3.488117218017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,1,power_law_1.2,0.3295808076858521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,1,power_law_1.2,0.3580096006393433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,1,balanced,5.656698862711589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,1,balanced,10.88039525349935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,1,power_law_1.2,0.43804159164428713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,1,power_law_1.2,0.4760128021240234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,1,balanced,7.794213612874349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,1,power_law_1.2,0.5171008110046387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,1,power_law_1.2,0.5807551860809326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,1,balanced,0.28942932685216266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,1,balanced,0.2890026569366455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,1,balanced,14.956175486246744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,1,power_law_1.01,2.6324159622192385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,1,power_law_1.2,0.7216320037841797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,1,power_law_1.01,0.27591040134429934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,1,power_law_1.01,0.2781055927276611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,1,power_law_1.01,1.5417856216430663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,1,power_law_1.2,0.8619327545166016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,1,power_law_1.01,1.9871807098388672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,1,power_law_1.2,2.6601728439331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,1,power_law_1.2,3.4473663330078126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,1,power_law_1.2,1.0734911918640138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,1,power_law_1.01,3.4187393188476562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,1,balanced,0.305456002553304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,1,balanced,0.30906132857004803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,1,balanced,0.3104533354441325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,1,power_law_1.01,0.2932415962219238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,1,power_law_1.01,0.3054080009460449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,1,power_law_1.01,0.33057920932769774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,1,power_law_1.01,2.898412895202637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,1,power_law_1.2,1.51147518157959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,1,power_law_1.2,0.057055997848510745
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,1,power_law_1.2,0.07571200132369996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,1,power_law_1.2,0.1074112057685852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,1,power_law_1.2,0.1368127942085266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,1,power_law_1.2,0.1408959984779358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,1,power_law_1.01,5.027731323242188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,1,power_law_1.2,0.16119680404663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,1,power_law_1.2,0.17997440099716186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,1,power_law_1.2,0.2009727954864502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,1,power_law_1.2,5.160351943969727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,1,power_law_1.2,0.17837439775466918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,1,power_law_1.2,0.18756480216979982
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,1,power_law_1.2,0.2106623888015747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,1,power_law_1.2,0.1942720055580139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,1,power_law_1.2,0.2031935930252075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,1,power_law_1.2,0.23164799213409423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,1,power_law_1.2,0.21825919151306153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,1,power_law_1.2,0.22935678958892822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,1,balanced,0.4477173487345378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,1,power_law_1.2,0.26677761077880857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,1,balanced,0.4542400042215983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,1,balanced,0.45899732907613117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,1,balanced,0.46647465229034424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,1,balanced,0.4774399995803833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,1,balanced,0.49479464689890545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,1,power_law_1.2,0.3139904022216797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,1,balanced,0.8056586583455404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,1,balanced,0.826848030090332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,1,balanced,0.8422079881032308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,1,balanced,1.2017707029978435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,1,power_law_1.01,3.810758590698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,1,balanced,1.5508480072021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,1,power_law_1.2,0.38093440532684325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,1,power_law_1.01,0.4493760108947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,1,power_law_1.2,0.44376959800720217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,1,power_law_1.2,1.9655872344970704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,1,power_law_1.01,0.48238720893859866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,1,power_law_1.01,0.5172416210174561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,1,power_law_1.01,0.6311423778533936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,1,power_law_1.01,0.6505280017852784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,1,power_law_1.01,0.7121664047241211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,1,balanced,2.2625706990559897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,1,balanced,2.874938646952311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,1,power_law_1.01,0.7616191864013672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,1,power_law_1.01,0.8328448295593261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,1,balanced,4.271914800008138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,1,power_law_1.01,1.057203197479248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,1,power_law_1.2,0.5047679901123047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,1,balanced,5.528554916381836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,1,power_law_1.01,1.404588794708252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,1,power_law_1.01,1.6876352310180665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,1,balanced,8.255013147989908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,1,power_law_1.01,2.3382144927978517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,1,balanced,10.576186498006185
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,1,power_law_1.2,0.7029759883880615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,1,balanced,16.13172785441081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,1,power_law_1.2,2.7724096298217775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,1,power_law_1.01,3.0149887084960936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,1,power_law_1.01,4.357510375976562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,1,balanced,21.168346405029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,1,power_law_1.01,5.714182281494141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,1,power_law_1.01,6.78344955444336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,1,power_law_1.2,0.9631999969482422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,1,power_law_1.2,7.232672119140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,1,balanced,0.2180160085360209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,1,power_law_1.01,8.359040069580079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,1,power_law_1.01,5.679654312133789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,1,power_law_1.2,3.769875335693359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,1,balanced,0.37161068121592206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,1,balanced,0.7158453464508057
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,1,power_law_1.2,1.2988672256469727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,1,balanced,0.7165706952412924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,1,balanced,0.719317356745402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,1,power_law_1.2,0.13537919521331787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,1,balanced,0.723578691482544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,1,power_law_1.01,11.156582641601563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,1,power_law_1.01,10.965529632568359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,1,balanced,0.7883093357086182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,1,balanced,0.7899413108825684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,1,power_law_1.2,0.17484159469604493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,1,power_law_1.2,0.1814784049987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,1,power_law_1.2,5.5745281219482425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,1,power_law_1.2,1.6909887313842773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,1,balanced,0.858890692392985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,1,balanced,0.8624053001403809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,1,power_law_1.2,11.13009262084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,1,power_law_1.01,7.882540893554688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,1,balanced,0.16127999623616537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,1,balanced,0.2524159948031108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,1,balanced,0.4352853298187256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,1,balanced,0.4378559986750285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,1,balanced,0.4362666606903076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,1,balanced,0.44686933358510333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,1,balanced,0.49081599712371826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,1,balanced,0.4936586618423462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,1,balanced,0.5196319818496704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,1,balanced,0.5178399880727133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,1,balanced,0.4291786750157674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,1,balanced,0.43617065747578937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,1,balanced,0.4438399871190389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,1,balanced,0.45741331577301025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,1,power_law_1.2,0.2780735969543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,1,balanced,0.6817386945088705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,1,power_law_1.2,0.2833856105804443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,1,balanced,0.693733294804891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,1,balanced,0.7097386519114176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,1,balanced,0.9334239959716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,1,balanced,0.9796853065490723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,1,balanced,1.2324000199635823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,1,balanced,1.5836639404296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,1,balanced,2.268821398417155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,1,balanced,3.031914710998535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,1,balanced,4.415402730305989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,1,power_law_1.01,16.271519470214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,1,balanced,0.7854719956715902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,1,balanced,5.857445398966472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,1,balanced,9.565568288167318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,1,power_law_1.2,0.29868159294128416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,1,power_law_1.2,0.3116863965988159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,1,power_law_1.2,0.334118390083313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,1,balanced,12.767232259114584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,1,power_law_1.2,7.9006591796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,1,balanced,0.7979733149210612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,1,balanced,0.7992853323618571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,1,balanced,0.8231200377146403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,1,balanced,1.3045066992441814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,1,balanced,1.3113120396931965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,1,balanced,1.332634687423706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,1,balanced,1.7253012657165527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,1,power_law_1.2,2.475935935974121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,1,balanced,1.7717119852701824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,1,power_law_1.2,0.44872322082519533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,1,power_law_1.2,0.4528319835662842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,1,power_law_1.2,0.46190080642700193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,1,power_law_1.2,0.6417664051055908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,1,power_law_1.2,0.6542335987091065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,1,power_law_1.2,0.7011712074279786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,1,power_law_1.2,0.7629759788513184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,1,power_law_1.2,0.8354944229125977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,1,balanced,0.0763626645008723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,1,balanced,0.1267039974530538
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,1,balanced,0.21802133321762085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,1,balanced,0.21975467602411905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,1,balanced,0.22271466255187988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,1,balanced,0.22378132740656534
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,1,balanced,0.35623466968536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,1,balanced,0.36061867078145343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,1,balanced,0.35946134726206463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,1,power_law_1.2,1.0374591827392579
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,1,balanced,0.3622026840845744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,1,balanced,0.36321600278218585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,1,balanced,0.37903467814127606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,1,balanced,0.38308799266815186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,1,balanced,0.3887360095977783
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,1,balanced,0.5339413483937582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,1,balanced,0.4649706681569417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,1,balanced,0.42643733819325763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,1,balanced,0.7897760073343912
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,1,balanced,0.45608532428741455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,1,power_law_1.2,1.3553600311279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,1,balanced,0.6326133410135905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,1,balanced,0.8442826271057129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,1,balanced,1.2310240268707275
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,1,balanced,1.6316107114156086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,1,power_law_1.2,1.6595840454101562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,1,balanced,2.202122688293457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,1,balanced,2.434096018473307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,1,balanced,2.8426186243693032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,1,balanced,3.3563572565714517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,1,balanced,4.46671994527181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,1,power_law_1.01,15.080403137207032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,1,balanced,5.10319455464681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,1,power_law_1.01,0.24298880100250245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,1,power_law_1.2,2.3582592010498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,1,power_law_1.01,0.34835200309753417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,1,power_law_1.01,0.44763522148132323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,1,balanced,5.6393280029296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,1,power_law_1.01,0.7232063770294189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,1,power_law_1.01,0.6945280075073242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,1,power_law_1.01,0.7723072052001954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,1,balanced,7.475189208984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,1,power_law_1.01,0.7894976139068604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,1,power_law_1.01,0.8637951850891114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,1,power_law_1.01,0.18491519689559938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,1,power_law_1.01,0.26056320667266847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,1,power_law_1.01,0.9195520401000976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,1,power_law_1.01,0.31004159450531005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,1,balanced,8.719088236490885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,1,power_law_1.01,0.41936001777648924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,1,power_law_1.01,1.024614429473877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,1,power_law_1.2,3.0528064727783204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,1,power_law_1.01,0.41911678314208983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,1,power_law_1.01,0.45613441467285154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,1,power_law_1.01,0.890726375579834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,1,power_law_1.01,0.5075071811676025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,1,power_law_1.01,0.9798272132873536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,1,power_law_1.01,0.5450687885284424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,1,power_law_1.01,0.5706624031066895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,1,power_law_1.01,1.0222016334533692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,1,power_law_1.01,0.6147007942199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,1,power_law_1.01,1.1392383575439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,1,power_law_1.01,0.4963840007781982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,1,power_law_1.01,0.5421887874603272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,1,power_law_1.01,1.2822463989257813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,1,balanced,11.929028828938803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,1,power_law_1.01,0.5566527843475342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,1,power_law_1.01,1.3568703651428222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,1,power_law_1.01,0.6044479846954346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,1,power_law_1.2,4.35588493347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,1,power_law_1.01,0.6860799789428711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,1,power_law_1.01,1.6652992248535157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,1,power_law_1.01,0.7146240234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,1,power_law_1.01,1.7720256805419923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,1,power_law_1.01,0.8661824226379394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,1,power_law_1.01,21.55677490234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,1,power_law_1.01,0.9450816154479981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,1,power_law_1.01,2.2469247817993163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,1,power_law_1.01,1.2054207801818848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,1,power_law_1.01,0.17749119997024537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,1,power_law_1.01,0.2495487928390503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,1,power_law_1.01,0.3001471996307373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,1,power_law_1.01,0.4299136161804199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,1,power_law_1.01,0.43987197875976564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,1,power_law_1.01,0.4624639987945557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,1,power_law_1.01,1.3623040199279786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,1,power_law_1.01,0.4787327766418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,1,power_law_1.01,2.4727359771728517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,1,balanced,19.086315155029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,1,power_law_1.01,0.5449984073638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,1,power_law_1.01,0.7677375793457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,1,power_law_1.01,0.7965312004089355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,1,power_law_1.01,0.8307647705078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,1,power_law_1.2,15.025439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,1,power_law_1.01,0.8963199615478515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,1,power_law_1.01,0.9502335548400879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,1,power_law_1.01,1.0524736404418946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,1,power_law_1.01,1.6922496795654296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,1,power_law_1.01,1.238700771331787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,1,power_law_1.01,1.3322303771972657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,1,power_law_1.01,1.6473535537719726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,1,power_law_1.01,3.1788415908813477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,1,power_law_1.01,2.158086395263672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,1,power_law_1.2,5.694220733642578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,1,balanced,25.699968973795574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,1,power_law_1.01,2.346579170227051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,1,power_law_1.01,2.660505676269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,1,power_law_1.2,0.21726078987121583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,1,power_law_1.2,0.31201920509338377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,1,power_law_1.2,0.398419189453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,1,power_law_1.01,3.7851329803466798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,1,power_law_1.2,0.7144576072692871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,1,power_law_1.01,4.299385452270508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,1,power_law_1.2,0.7026559829711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,1,power_law_1.2,0.7965375900268554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,1,power_law_1.01,3.067820739746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,1,power_law_1.2,0.8026559829711915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,1,power_law_1.01,4.851724624633789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,1,power_law_1.2,0.8674816131591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,1,power_law_1.2,0.9224703788757325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,1,power_law_1.2,0.9580224037170411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,1,power_law_1.2,0.9180095672607422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,1,power_law_1.2,1.0011648178100585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,1,power_law_1.2,1.0317760467529298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,1,power_law_1.2,1.1636480331420898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,1,power_law_1.2,1.2051712036132813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,1,power_law_1.2,1.3461312294006347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,1,power_law_1.01,7.0150398254394535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,1,power_law_1.2,1.6390016555786133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,1,power_law_1.2,1.8283327102661133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,1,power_law_1.01,0.07859200239181519
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,1,power_law_1.01,0.11214079856872558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,1,power_law_1.01,5.734783935546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,1,power_law_1.01,0.14684159755706788
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,1,power_law_1.01,0.2156735897064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,1,power_law_1.01,0.22621440887451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,1,power_law_1.01,0.2930880069732666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,1,power_law_1.01,4.646527862548828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,1,power_law_1.01,0.348799991607666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,1,power_law_1.01,0.41070079803466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,1,power_law_1.2,2.207014465332031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,1,power_law_1.01,0.3258752107620239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,1,power_law_1.01,0.368121600151062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,1,power_law_1.01,0.42807679176330565
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,1,power_law_1.01,0.3465856075286865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,1,power_law_1.01,0.3849152088165283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,1,power_law_1.2,8.357945251464844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,1,power_law_1.01,0.4579328060150146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,1,power_law_1.01,0.3933183908462524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,1,power_law_1.01,0.42806401252746584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,1,power_law_1.01,0.49410557746887207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,1,power_law_1.2,2.438470458984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,1,power_law_1.01,0.4592832088470459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,1,power_law_1.01,8.939871978759765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,1,power_law_1.2,3.3055999755859373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,1,power_law_1.2,0.14991999864578248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,1,power_law_1.2,0.18915200233459473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,1,power_law_1.01,0.6211584091186524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,1,power_law_1.2,0.26629760265350344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,1,power_law_1.01,9.15411834716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,1,power_law_1.2,0.42513279914855956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,1,power_law_1.2,0.4014400005340576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,1,power_law_1.2,0.46262397766113283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,1,power_law_1.2,0.5099904060363769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,1,power_law_1.2,0.5676224231719971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,1,power_law_1.2,0.5720128059387207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,1,power_law_1.01,0.7721151828765869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,1,power_law_1.2,0.6008128166198731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,1,power_law_1.2,3.079507255554199
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,1,power_law_1.01,1.0038080215454102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,1,power_law_1.2,0.48215680122375487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,1,power_law_1.2,0.5455743789672851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,1,power_law_1.2,0.5529471874237061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,1,power_law_1.2,0.5954751968383789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,1,power_law_1.2,0.6494847774505615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,1,power_law_1.2,0.707583999633789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,1,power_law_1.2,0.8696703910827637
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,1,power_law_1.01,1.3733887672424316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,1,power_law_1.2,0.964748764038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,1,power_law_1.01,5.75225601196289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,1,power_law_1.2,1.188377571105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,1,power_law_1.01,13.368684387207031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,1,power_law_1.2,1.3369024276733399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,1,power_law_1.2,4.435078430175781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,1,power_law_1.2,10.950521850585938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,1,power_law_1.2,1.6833919525146483
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,1,power_law_1.01,1.8015680313110352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,1,power_law_1.01,12.183904266357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,1,power_law_1.2,5.754726409912109
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,1,power_law_1.01,2.608652877807617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,1,power_law_1.2,2.356038475036621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,1,balanced,0.1796906590461731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,1,power_law_1.01,17.71233215332031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,1,power_law_1.2,3.0729536056518554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,1,power_law_1.01,9.457459259033204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,1,power_law_1.2,8.99151382446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,1,balanced,0.2653706669807434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,1,balanced,0.4464799960454305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,1,power_law_1.2,4.533369445800782
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,1,power_law_1.01,3.4235904693603514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,1,power_law_1.2,12.015245056152343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,1,balanced,0.4491093158721924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,1,balanced,0.4505973259607951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,1,power_law_1.2,16.20394287109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,1,power_law_1.01,18.581114196777342
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,1,power_law_1.01,5.018380737304687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,1,power_law_1.2,6.180198287963867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,1,balanced,0.46061333020528156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,1,balanced,0.465338667233785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,1,balanced,0.46767465273539227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,1,power_law_1.01,26.217791748046874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,1,power_law_1.01,12.8708740234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,1,power_law_1.2,18.306521606445312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,1,balanced,0.7621599833170573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,1,balanced,0.7651360034942627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,1,balanced,0.7716000080108643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,1,balanced,0.7835520108540853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,1,balanced,0.7945653597513834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,1,balanced,0.8159733613332113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,1,balanced,1.28001602490743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,1,balanced,1.3001759847005208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,1,balanced,1.325434684753418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,1,balanced,1.8544692993164062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,1,balanced,2.4813386599222818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,1,balanced,3.543013254801432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,1,balanced,4.665727933247884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,1,balanced,6.844906489054362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,1,balanced,8.930741628011068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,1,balanced,13.437012990315756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,1,balanced,17.258890787760418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,1,power_law_1.01,24.798136901855468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,1,balanced,26.15167490641276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,1,power_law_1.01,6.7559356689453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,1,power_law_1.2,9.477049255371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,1,balanced,34.45447540283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,1,balanced,0.14426133036613464
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,1,power_law_1.2,0.0760703980922699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,1,power_law_1.2,0.10527360439300537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,1,power_law_1.01,34.80891418457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,1,power_law_1.2,0.12791680097579955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,1,power_law_1.2,0.2107072114944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,1,power_law_1.2,0.23121280670166017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,1,balanced,0.08457600076993306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,1,power_law_1.2,0.2827647924423218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,1,power_law_1.2,0.3389695882797241
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,1,power_law_1.2,0.4199552059173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,1,balanced,0.2425866723060608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,1,balanced,0.42710399627685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,1,power_law_1.2,0.32607359886169435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,1,balanced,0.8242506980895996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,1,power_law_1.2,0.3506175994873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,1,power_law_1.2,0.4192319869995117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,1,power_law_1.2,0.36039679050445556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,1,power_law_1.2,0.37329280376434326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,1,power_law_1.2,0.4480576038360596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,1,power_law_1.2,0.39635839462280276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,1,power_law_1.2,0.4164608001708984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,1,power_law_1.2,0.5086207866668702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,1,power_law_1.2,0.49417600631713865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,1,power_law_1.2,21.52684783935547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,1,balanced,0.1439306636651357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,1,balanced,0.2537226676940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,1,balanced,0.4749386707941691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,1,power_law_1.2,0.5616127967834472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,1,balanced,1.5980587005615234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,1,balanced,0.09674132863680522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,1,balanced,3.145344098409017
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,1,power_law_1.2,0.7517824172973633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,1,balanced,0.05845333139101664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,1,balanced,0.08644266923268636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,1,balanced,0.14074132839838663
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,1,balanced,0.24675200382868448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,1,balanced,0.4614880084991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,1,balanced,0.889952023824056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,1,power_law_1.2,1.0204607963562011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,1,balanced,0.9002506732940674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,1,balanced,0.9042719999949137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,1,balanced,0.9147093296051025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,1,balanced,3.254805246988932
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,1,balanced,0.908560037612915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,1,balanced,1.797152042388916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,1,balanced,0.9111413160959879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,1,balanced,0.9197493394215902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,1,balanced,3.3005174001057944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,1,balanced,0.15056533614794412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,1,balanced,0.2646080056826274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,1,balanced,0.50109334786733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,1,balanced,0.9279092947642008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,1,balanced,0.9395039876302084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,1,balanced,0.954906702041626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,1,balanced,1.0446080366770427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,1,balanced,1.0594933032989502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,1,balanced,1.0908532937367756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,1,balanced,1.0984533627827961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,1,balanced,1.1396479606628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,1,balanced,1.2431306838989258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,1,balanced,1.3151466846466064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,1,balanced,1.8368159929911296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,1,balanced,1.65665070215861
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,1,balanced,3.3168532053629556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,1,balanced,2.201845328013102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,1,balanced,2.0088000297546387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,1,balanced,3.6320587793986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,1,balanced,3.2546666463216147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,1,balanced,2.009215990702311
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,1,power_law_1.2,1.4112192153930665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,1,balanced,3.6536267598470054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,1,balanced,4.169887860616048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,1,balanced,0.9380799929300944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,1,power_law_1.2,25.817156982421874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,1,balanced,1.911194642384847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,1,balanced,1.9168319702148438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,1,balanced,1.826090653737386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,1,balanced,1.8336373964945476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,1,power_law_1.01,0.145907199382782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,1,balanced,3.472874641418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,1,power_law_1.2,0.17679359912872314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,1,balanced,1.8391413688659668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,1,power_law_1.2,1.824448013305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,1,power_law_1.01,0.2385279893875122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,1,power_law_1.01,0.08171520233154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,1,power_law_1.01,0.422438383102417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,1,power_law_1.01,0.8183039665222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,1,balanced,3.468245188395182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,1,balanced,3.5820960998535156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,1,power_law_1.2,2.6101375579833985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,1,balanced,1.778864065806071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,1,balanced,3.6448853810628257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,1,balanced,3.6475998560587564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,1,balanced,3.641551971435547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,1,power_law_1.2,0.24546558856964112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,1,balanced,3.712378819783529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,1,power_law_1.2,0.31722240447998046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,1,power_law_1.01,0.14248960018157958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,1,power_law_1.01,0.25525119304656985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,1,power_law_1.01,0.4712831974029541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,1,power_law_1.01,1.260204792022705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,1,balanced,2.94538148244222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,1,power_law_1.2,12.675174713134766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,1,power_law_1.01,1.6372095108032227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,1,balanced,1.7996320724487305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,1,balanced,1.8057173093159993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,1,balanced,2.954373359680176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,1,balanced,1.8129067420959473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,1,balanced,1.8392213185628254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,1,balanced,2.967349370320638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,1,balanced,1.8488640785217285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,1,balanced,2.97923215230306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,1,balanced,1.8938825925191243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,1,power_law_1.2,0.43614721298217773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,1,power_law_1.2,0.4451456069946289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,1,balanced,2.9933547973632812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,1,balanced,4.06438414255778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,1,balanced,3.0114399592081704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,1,balanced,3.033381462097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,1,balanced,4.359813372294108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,1,power_law_1.01,0.6875391960144043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,1,power_law_1.01,0.9117312431335449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,1,balanced,3.05619208017985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,1,balanced,4.547962824503581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,1,balanced,3.0932534535725913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,1,balanced,4.570528030395508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,1,balanced,3.1614507039388022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,1,power_law_1.01,2.38657283782959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,1,balanced,3.216986656188965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,1,balanced,4.720858573913574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,1,power_law_1.01,2.396774482727051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,1,balanced,3.3237279256184897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,1,balanced,4.911706606547038
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,1,power_law_1.2,3.432979202270508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,1,balanced,3.449199994405111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,1,balanced,5.919178644816081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,1,balanced,8.166831970214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,1,balanced,6.136816024780273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,1,power_law_1.2,0.44895358085632325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,1,power_law_1.2,0.482374382019043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,1,power_law_1.2,0.5381311893463134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,1,power_law_1.01,1.4502911567687988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,1,balanced,8.819658915201822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,1,balanced,2.1991626421610513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,1,power_law_1.01,1.5334336280822753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,1,balanced,2.2270827293395996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,1,balanced,8.916725158691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,1,balanced,2.322901407877604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,1,balanced,2.5008533795674643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,1,balanced,2.6576053301493325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,1,balanced,12.143243153889975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,1,balanced,2.8179146448771157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,1,power_law_1.01,2.8291839599609374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,1,balanced,11.631295522054037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,1,balanced,4.424634615580241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,1,power_law_1.01,2.9288703918457033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,1,balanced,4.718400001525879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,1,balanced,16.092875162760418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,1,balanced,6.8573869069417315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,1,balanced,17.0916748046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,1,balanced,9.187290827433268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,1,power_law_1.01,1.5185664176940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,1,power_law_1.01,1.5604479789733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,1,balanced,22.483749389648438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,1,power_law_1.01,0.05720319747924805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,1,power_law_1.2,0.767142391204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,1,power_law_1.01,2.960767936706543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,1,power_law_1.01,0.0849727988243103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,1,power_law_1.2,0.7795072078704834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,1,power_law_1.01,0.13859200477600098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,1,power_law_1.2,0.8282367706298828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,1,power_law_1.01,0.2466752052307129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,1,power_law_1.2,0.9155455589294433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,1,power_law_1.01,0.36100480556488035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,1,power_law_1.01,0.482860803604126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,1,power_law_1.2,0.9228608131408691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,1,power_law_1.01,0.6714240074157715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,1,power_law_1.2,1.0523903846740723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,1,power_law_1.01,0.7030399799346924
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,1,power_law_1.01,0.7232448101043701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,1,power_law_1.2,1.2284671783447265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,1,power_law_1.01,0.7380608081817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,1,power_law_1.2,1.317683219909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,1,power_law_1.01,0.7651328086853028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,1,power_law_1.2,0.14515199661254882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,1,power_law_1.01,0.8029248237609863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,1,power_law_1.2,1.6236608505249024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,1,power_law_1.01,0.8221695899963379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,1,power_law_1.01,0.9171199798583984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,1,power_law_1.01,0.9771264076232911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,1,power_law_1.2,2.162220764160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,1,power_law_1.01,1.059507179260254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,1,power_law_1.01,1.2175040245056152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,1,power_law_1.01,1.1422464370727539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,1,power_law_1.2,2.6704639434814452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,1,power_law_1.01,1.4670080184936523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,1,power_law_1.01,1.3509632110595704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,1,power_law_1.2,0.08022400140762329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,1,power_law_1.01,1.3433279991149902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,1,power_law_1.01,1.549567985534668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,1,power_law_1.01,2.876742362976074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,1,power_law_1.01,1.6022144317626954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,1,power_law_1.01,1.9038848876953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,1,power_law_1.01,3.0018367767333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,1,power_law_1.2,0.24029440879821778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,1,power_law_1.2,0.42429437637329104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,1,power_law_1.01,2.1909887313842775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,1,power_law_1.01,3.042348861694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,1,power_law_1.2,3.805414581298828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,1,power_law_1.01,2.6402944564819335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,1,power_law_1.01,3.1296575546264647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,1,power_law_1.2,0.8224960327148437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,1,power_law_1.2,4.842342376708984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,1,power_law_1.01,3.193312072753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,1,power_law_1.01,3.542758560180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,1,power_law_1.2,0.1478592038154602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,1,power_law_1.01,3.291455841064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,1,power_law_1.2,0.2567296028137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,1,power_law_1.2,0.462713623046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,1,power_law_1.01,5.504415893554688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,1,power_law_1.01,1.504032039642334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,1,power_law_1.01,1.5453248023986816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,1,power_law_1.01,1.5903424263000487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,1,power_law_1.01,1.6017023086547852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,1,power_law_1.01,1.663212776184082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,1,power_law_1.01,1.7059711456298827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,1,power_law_1.2,7.037100982666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,1,power_law_1.2,1.1154303550720215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,1,power_law_1.2,5.12603530883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,1,power_law_1.2,1.4849599838256835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,1,power_law_1.01,3.9082046508789063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,1,power_law_1.2,0.6191616058349609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,1,power_law_1.2,0.8546239852905273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,1,power_law_1.01,4.1260734558105465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,1,power_law_1.01,4.606195068359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,1,power_law_1.01,5.00561294555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,1,power_law_1.01,2.021561622619629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,1,power_law_1.01,2.154764747619629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,1,power_law_1.01,5.96563835144043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,1,power_law_1.2,2.345439910888672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,1,power_law_1.01,2.425388717651367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,1,power_law_1.2,2.4710655212402344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,1,power_law_1.01,2.5802047729492186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,1,power_law_1.2,1.3986432075500488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,1,power_law_1.01,7.080461120605468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,1,power_law_1.01,3.121958351135254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,1,power_law_1.2,1.5297087669372558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,1,power_law_1.01,3.6499649047851563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,1,power_law_1.01,8.520684814453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,1,power_law_1.01,4.738598251342774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,1,power_law_1.01,5.592416000366211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,1,power_law_1.01,10.764723205566407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,1,power_law_1.01,7.672697448730469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,1,power_law_1.2,2.841209602355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,1,power_law_1.2,9.086227416992188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,1,power_law_1.2,2.8047040939331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,1,power_law_1.01,14.027110290527343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,1,power_law_1.01,9.859731292724609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,1,power_law_1.2,1.50315523147583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,1,power_law_1.2,1.5274432182312012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,1,power_law_1.01,17.659596252441407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,1,power_law_1.01,0.09438080191612244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,1,balanced,0.1458239952723185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,1,balanced,0.24475200970967612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,1,balanced,0.4079573154449463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,1,balanced,0.7234079837799072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,1,balanced,1.394752025604248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,1,balanced,2.701109250386556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,1,balanced,2.8347574869791665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,1,balanced,2.9898398717244468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,1,balanced,3.155674616495768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,1,power_law_1.2,2.826579284667969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,1,balanced,3.155167897542318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,1,power_law_1.2,0.05724160075187683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,1,power_law_1.2,0.08516479730606079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,1,balanced,3.0742400487264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,1,power_law_1.2,0.13838720321655273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,1,power_law_1.2,0.23621759414672852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,1,balanced,3.1048641204833984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,1,power_law_1.2,0.32524800300598145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,1,power_law_1.2,0.44985599517822267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,1,balanced,3.1046667098999023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,1,power_law_1.2,13.377574157714843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,1,balanced,3.129530588785807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,1,power_law_1.2,1.4535743713378906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,1,balanced,3.183919906616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,1,power_law_1.01,0.14884480237960815
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,1,power_law_1.2,0.6531712055206299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,1,balanced,3.203920046488444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,1,power_law_1.01,0.2663680076599121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,1,power_law_1.01,0.49987201690673827
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,1,power_law_1.2,0.6849664211273193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,1,balanced,3.2711305618286133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,1,power_law_1.2,0.7046080112457276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,1,power_law_1.2,0.7448895931243896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,1,balanced,3.845786730448405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,1,power_law_1.2,0.7712512016296387
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,1,power_law_1.2,0.7958784103393555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,1,power_law_1.2,0.8542207717895508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,1,balanced,3.836186726888021
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,1,power_law_1.2,0.9257599830627441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,1,power_law_1.2,1.0223615646362305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,1,balanced,3.900005340576172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,1,power_law_1.2,1.107744026184082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,1,balanced,4.003066698710124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,1,power_law_1.2,2.884896087646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,1,power_law_1.2,1.2954367637634276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,1,power_law_1.2,1.2140543937683106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,1,balanced,4.22378667195638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,1,power_law_1.2,2.9530176162719726
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,1,power_law_1.2,1.3842623710632325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,1,power_law_1.2,1.4014399528503418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,1,balanced,4.420218785603841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,1,power_law_1.2,3.1351551055908202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,1,power_law_1.2,1.627065658569336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,1,power_law_1.2,3.141209602355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,1,power_law_1.2,1.661734390258789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,1,balanced,6.886714935302734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,1,power_law_1.2,1.5047167778015136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,1,power_law_1.2,1.939129638671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,1,power_law_1.2,3.2632190704345705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,1,power_law_1.2,1.5264639854431152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,1,power_law_1.2,2.2859519958496093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,1,power_law_1.2,1.5604800224304198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,1,balanced,7.447760264078776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,1,power_law_1.2,3.467295837402344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,1,power_law_1.2,1.6275007247924804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,1,power_law_1.2,2.7312639236450194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,1,power_law_1.2,1.70446720123291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,1,power_law_1.2,1.7986560821533204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,1,power_law_1.2,3.6272449493408203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,1,power_law_1.01,0.7360576152801513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,1,balanced,10.62784512837728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,1,power_law_1.2,5.715110397338867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,1,balanced,13.60421371459961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,1,balanced,0.0765173335870107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,1,balanced,0.13030399878819784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,1,balanced,0.23109867175420126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,1,balanced,0.4137173493703206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,1,balanced,0.7856533527374268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,1,balanced,1.540170669555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,1,balanced,1.7336959838867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,1,balanced,1.794101397196452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,1,balanced,1.670346736907959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,1,balanced,1.7293705940246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,1,balanced,1.5880640347798665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,1,balanced,1.5895519256591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,1,balanced,1.586880048116048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,1,balanced,1.595450719197591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,1,power_law_1.2,3.989510345458984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,1,balanced,1.6141386032104492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,1,power_law_1.2,17.655474853515624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,1,balanced,1.6207839647928874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,1,power_law_1.01,0.9542079925537109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,1,balanced,1.6285707155863445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,1,power_law_1.2,4.356486511230469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,1,power_law_1.01,1.3556223869323731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,1,power_law_1.2,2.1375871658325196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,1,power_law_1.01,1.3836416244506835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,1,power_law_1.2,2.2481088638305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,1,power_law_1.2,4.7475841522216795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,1,power_law_1.2,2.5420480728149415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,1,balanced,2.0012906392415366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,1,balanced,1.9292532602945964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,1,balanced,0.08638933300971985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,1,power_law_1.2,2.7793407440185547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,1,balanced,2.001248041788737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,1,power_law_1.2,5.129625701904297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,1,balanced,2.121583938598633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,1,power_law_1.2,3.200806427001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,1,balanced,2.3335092862447104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,1,balanced,2.4728426933288574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,1,power_law_1.2,6.266963195800781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,1,power_law_1.2,3.6460033416748048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,1,balanced,3.765455881754557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,1,balanced,4.404725392659505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,1,power_law_1.2,4.740972900390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,1,power_law_1.2,7.161587524414062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,1,power_law_1.2,0.09587200284004212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,1,balanced,5.740506490071614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,1,power_law_1.2,5.521382522583008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,1,power_law_1.2,6.821920013427734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,1,power_law_1.2,8.872921752929688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,1,balanced,7.613077163696289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,1,power_law_1.2,8.016921234130859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,1,balanced,0.1339359978834788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,1,balanced,0.23323200146357217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,1,balanced,0.43274664878845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,1,power_law_1.2,10.808601379394531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,1,power_law_1.01,0.12524160146713256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,1,power_law_1.2,9.919910430908203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,1,power_law_1.01,0.0744383990764618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,1,power_law_1.01,0.13296639919281006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,1,power_law_1.01,0.22791039943695068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,1,power_law_1.01,0.410368013381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,1,power_law_1.01,0.6017216205596924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,1,power_law_1.01,0.8095295906066895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,1,balanced,0.0524533341328303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,1,power_law_1.2,14.530546569824219
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,1,balanced,0.0769706666469574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,1,balanced,0.1220266620318095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,1,power_law_1.2,0.14878720045089722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,1,power_law_1.01,2.351910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,1,balanced,0.2160266637802124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,1,balanced,0.4001386562983195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,1,power_law_1.2,0.26721279621124266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,1,power_law_1.01,1.2496959686279296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,1,balanced,0.7671733697255453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,1,balanced,0.7726186911265055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,1,power_law_1.2,0.4816768169403076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,1,balanced,0.7806666692097982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,1,power_law_1.01,1.309331226348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,1,power_law_1.01,2.445568084716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,1,balanced,0.7812159856160482
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,1,balanced,0.7844106356302897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,1,power_law_1.01,1.279750442504883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,1,balanced,0.7934026718139648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,1,balanced,0.7997066974639893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,1,power_law_1.01,2.4843008041381838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,1,balanced,0.8082239627838135
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,1,balanced,0.8235946496327718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,1,balanced,0.8966613610585531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,1,power_law_1.01,2.540934371948242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,1,balanced,0.9085386594136556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,1,balanced,0.9347679615020752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,1,balanced,0.8044160207112631
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,1,balanced,0.942629337310791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,1,power_law_1.01,0.21074559688568115
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,1,balanced,0.9784533182779948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,1,power_law_1.01,2.562086486816406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,1,power_law_1.01,0.3681344032287598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,1,power_law_1.01,1.296121597290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,1,balanced,1.0634026527404785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,1,power_law_1.01,0.7084479808807373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,1,balanced,1.121519962946574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,1,power_law_1.01,1.2698240280151367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,1,power_law_1.01,2.598271942138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,1,balanced,1.538954734802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,1,power_law_1.2,17.545689392089844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,1,power_law_1.01,1.3052800178527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,1,balanced,1.3880693117777507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,1,power_law_1.01,1.3228032112121582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,1,power_law_1.01,2.6868480682373046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,1,balanced,2.8438825607299805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,1,power_law_1.01,1.3527487754821776
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,1,balanced,1.8896907170613606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,1,power_law_1.01,2.7720447540283204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,1,power_law_1.01,1.370860767364502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,1,balanced,2.7591253916422525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,1,power_law_1.01,1.4133184432983399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,1,power_law_1.01,2.967091178894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,1,power_law_1.01,1.4935744285583497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,1,balanced,3.6963574091593423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,1,power_law_1.01,1.762303924560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,1,power_law_1.01,3.2492416381835936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,1,power_law_1.01,1.8924287796020507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,1,power_law_1.01,3.6083454132080077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,1,power_law_1.01,2.062233543395996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,1,power_law_1.01,2.238662338256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,1,power_law_1.01,4.24186897277832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,1,power_law_1.01,1.0530943870544434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,1,power_law_1.01,2.7510080337524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,1,power_law_1.01,1.441875171661377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,1,power_law_1.01,3.081772804260254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,1,power_law_1.01,4.945568084716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,1,balanced,1.5614879926045735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,1,power_law_1.2,0.6552576065063477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,1,balanced,1.5671359697977703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,1,balanced,1.5976373354593914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,1,power_law_1.01,0.08521599769592285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,1,power_law_1.01,3.7982719421386717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,1,power_law_1.01,6.199513626098633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,1,power_law_1.01,4.849555206298828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,1,power_law_1.01,0.05080320239067078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,1,power_law_1.01,0.07448319792747497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,1,power_law_1.01,0.1217919945716858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,1,power_law_1.01,0.2144767999649048
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,1,power_law_1.01,0.3130239963531494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,1,power_law_1.01,7.590656280517578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,1,power_law_1.01,0.42405118942260744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,1,power_law_1.01,6.903616333007813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,1,power_law_1.01,0.5788159847259522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,1,power_law_1.01,0.5972224235534668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,1,power_law_1.01,0.6227392196655274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,1,power_law_1.01,0.6373055934906006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,1,power_law_1.01,0.6651072025299072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,1,power_law_1.01,0.6909440040588379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,1,power_law_1.01,2.04256649017334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,1,power_law_1.01,10.234028625488282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,1,power_law_1.01,0.7078144073486328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,1,power_law_1.01,8.268460845947265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,1,power_law_1.01,0.7844672203063965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,1,power_law_1.01,2.154092788696289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,1,power_law_1.01,0.8326144218444824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,1,power_law_1.01,0.9030655860900879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,1,power_law_1.01,1.0541440010070802
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,1,power_law_1.01,0.9934016227722168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,1,power_law_1.2,0.8877311706542969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,1,power_law_1.01,0.1343168020248413
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,1,power_law_1.01,1.1492032051086425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,1,power_law_1.01,0.23192319869995118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,1,power_law_1.2,1.323526382446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,1,power_law_1.01,1.1446720123291017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,1,power_law_1.01,0.43357439041137696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,1,power_law_1.01,12.928274536132813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,1,power_law_1.01,1.3500608444213866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,1,power_law_1.2,1.3816512107849122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,1,power_law_1.01,1.3713855743408203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,1,power_law_1.01,1.618783950805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,1,power_law_1.01,1.870195198059082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,1,power_law_1.01,2.2549888610839846
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,1,power_law_1.01,3.004915237426758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,1,power_law_1.01,2.419270324707031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,1,power_law_1.01,18.139308166503906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,1,power_law_1.01,4.9485633850097654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,1,power_law_1.01,2.5617279052734374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,1,balanced,2.5265599886576333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,1,power_law_1.2,0.12762240171432496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,1,balanced,2.5348532994588218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,1,balanced,2.5450080235799155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,1,power_law_1.2,26.166778564453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,1,balanced,2.558634599049886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,1,balanced,2.57094399134318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,1,power_law_1.01,0.6383039951324463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,1,balanced,2.5891040166219077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,1,balanced,2.6088266372680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,1,power_law_1.2,0.07303680181503296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,1,balanced,2.6263039906819663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,1,balanced,2.6606879234313965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,1,balanced,2.72486941019694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,1,power_law_1.01,23.274981689453124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,1,balanced,2.7751251856486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,1,balanced,2.8766934076944985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,1,power_law_1.2,0.20992639064788818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,1,power_law_1.2,0.3664448022842407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,1,balanced,2.992682774861654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,1,power_law_1.2,0.7064640045166015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,1,balanced,5.1192372639973955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,1,balanced,5.319770812988281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,1,power_law_1.2,0.13031680583953859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,1,power_law_1.2,0.2260672092437744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,1,power_law_1.01,2.464064025878906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,1,power_law_1.2,0.39439361095428466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,1,balanced,7.727904001871745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,1,power_law_1.01,0.8416000366210937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,1,power_law_1.2,2.3424896240234374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,1,power_law_1.01,1.1421183586120605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,1,power_law_1.01,1.2144960403442382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,1,power_law_1.2,2.349420738220215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,1,power_law_1.2,2.4263423919677733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,1,balanced,10.087184270222982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,1,power_law_1.2,0.08477439880371093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,1,power_law_1.2,2.563302421569824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,1,power_law_1.2,2.525132751464844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,1,power_law_1.2,2.630169677734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,1,balanced,14.815930684407553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,1,power_law_1.2,0.9691583633422851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,1,power_law_1.2,0.5352960109710694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,1,power_law_1.2,2.7882944107055665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,1,power_law_1.2,0.7241216182708741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,1,power_law_1.2,1.2518400192260741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,1,power_law_1.2,2.866009521484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,1,power_law_1.2,0.13405439853668213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,1,power_law_1.2,0.23096320629119874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,1,power_law_1.2,3.083270454406738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,1,power_law_1.01,2.5319103240966796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,1,power_law_1.2,0.4145472049713135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,1,power_law_1.01,2.548543930053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,1,power_law_1.2,3.4768001556396486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,1,power_law_1.01,2.608172798156738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,1,balanced,19.480091094970703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,1,power_law_1.2,3.751417541503906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,1,power_law_1.01,2.714348793029785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,1,power_law_1.01,2.764499282836914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,1,power_law_1.2,4.38853759765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,1,power_law_1.01,2.8010496139526366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,1,power_law_1.2,0.5467904090881348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,1,power_law_1.2,5.099878311157227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,1,power_law_1.2,1.2577280044555663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,1,power_law_1.2,2.001363182067871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,1,power_law_1.2,1.2966336250305175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,1,power_law_1.2,2.115123176574707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,1,power_law_1.2,6.435391998291015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,1,power_law_1.2,0.051148802042007446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,1,power_law_1.2,0.07576320171356202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,1,power_law_1.2,0.12201600074768067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,1,power_law_1.01,2.041804885864258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,1,power_law_1.2,0.19876480102539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,1,power_law_1.2,7.736115264892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,1,power_law_1.2,0.28222079277038575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,1,power_law_1.2,0.7665023803710938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,1,power_law_1.2,0.3889472007751465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,1,power_law_1.2,1.1165184020996093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,1,power_law_1.2,0.5485184192657471
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,1,power_law_1.2,0.5979263782501221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,1,power_law_1.2,1.190073585510254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,1,power_law_1.2,0.6227968215942383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,1,power_law_1.2,0.6458879947662354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,1,power_law_1.01,2.099910354614258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,1,power_law_1.2,0.672435188293457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,1,power_law_1.2,0.6928768157958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,1,power_law_1.2,10.366822052001954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,1,power_law_1.01,2.1691904067993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,1,power_law_1.2,0.7227327823638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,1,power_law_1.2,0.8071359634399414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,1,power_law_1.01,2.152095985412598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,1,power_law_1.2,0.8700351715087891
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,1,power_law_1.2,0.9498368263244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,1,power_law_1.01,2.198259162902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,1,power_law_1.01,3.2968894958496096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,1,power_law_1.2,2.355718421936035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,1,power_law_1.2,1.1196288108825683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,1,power_law_1.01,2.315878486633301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,1,power_law_1.2,1.0398336410522462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,1,power_law_1.2,2.5152320861816406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,1,power_law_1.01,3.5875198364257814
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,1,power_law_1.2,1.2067456245422363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,1,power_law_1.2,1.293836784362793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,1,power_law_1.01,2.2732416152954102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,1,power_law_1.2,1.1923904418945312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,1,power_law_1.2,12.946214294433593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,1,power_law_1.2,1.2988096237182618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,1,power_law_1.01,2.360006332397461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,1,power_law_1.01,3.762860870361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,1,power_law_1.2,1.4019200325012207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,1,power_law_1.01,2.4916864395141602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,1,power_law_1.2,1.4282560348510742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,1,power_law_1.01,4.239673614501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,1,power_law_1.2,1.6745344161987306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,1,power_law_1.01,2.8022335052490233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,1,power_law_1.2,1.9888832092285156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,1,power_law_1.2,1.9634496688842773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,1,power_law_1.01,3.1487808227539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,1,power_law_1.01,4.9858753204345705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,1,power_law_1.2,2.021011161804199
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,1,power_law_1.2,2.3365631103515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,1,power_law_1.2,2.093964767456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,1,power_law_1.01,3.6257919311523437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,1,power_law_1.2,3.107923126220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,1,power_law_1.2,18.103692626953126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,1,power_law_1.2,2.145894432067871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,1,power_law_1.01,6.003763198852539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,1,power_law_1.01,4.228179168701172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,1,power_law_1.2,2.2338111877441404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,1,power_law_1.2,34.81514892578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,1,power_law_1.2,5.293926239013672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,1,power_law_1.2,2.2756160736083983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,1,power_law_1.01,7.539775848388672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,1,power_law_1.2,2.386790466308594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,1,power_law_1.2,2.4737216949462892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,1,power_law_1.01,5.404000091552734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,1,power_law_1.01,8.54233627319336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,1,power_law_1.2,2.611084747314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,1,power_law_1.2,1.2694463729858398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,1,power_law_1.2,2.9629823684692385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,1,balanced,0.025402667621771496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,1,power_law_1.01,11.559820556640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,1,power_law_1.2,3.2999744415283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,1,power_law_1.2,2.4435007095336916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,1,power_law_1.2,23.14970245361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,1,power_law_1.01,6.567533111572265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,1,balanced,0.029311999678611755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,1,power_law_1.2,3.8261566162109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,1,power_law_1.2,4.374361419677735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,1,power_law_1.01,8.951058959960937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,1,power_law_1.01,14.773464965820313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,1,power_law_1.2,5.568345642089843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,1,balanced,0.031583999594052635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,1,balanced,0.04993066688378652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,1,balanced,0.027434666951497395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,1,balanced,0.030207999050617218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,1,balanced,0.035504000882307686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,1,balanced,0.05545066793759664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,1,balanced,0.07904533545176189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,1,balanced,0.041509332756201424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,1,balanced,0.08057066798210144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,1,balanced,0.08345066507657369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,1,power_law_1.2,6.69164810180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,1,balanced,0.08596266309420268
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,1,balanced,0.08649067083994548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,1,balanced,0.08683199683825175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,1,balanced,0.08900266885757446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,1,balanced,0.09596799810727437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,1,balanced,0.09880533814430237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,1,balanced,0.10053333640098572
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,1,balanced,0.09833066662152608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,1,power_law_1.2,1.3018752098083497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,1,balanced,0.10172800223032634
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,1,balanced,0.10523733496665955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,1,balanced,0.11590933799743652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,1,balanced,0.12274666627248128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,1,balanced,0.16010666886965433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,1,balanced,0.15548266967137656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,1,power_law_1.2,1.3187840461730957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,1,balanced,0.29734400908152264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,1,balanced,0.24241065979003906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,1,balanced,0.3393706480662028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,1,power_law_1.01,11.195206451416016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,1,balanced,0.43145068486531574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,1,power_law_1.2,1.380083179473877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,1,balanced,0.6101973454157511
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,1,balanced,0.8028213183085123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,1,power_law_1.2,1.3904704093933105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,1,power_law_1.2,2.4886207580566406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,1,power_law_1.2,8.987686157226562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,1,power_law_1.2,1.4686911582946778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,1,power_law_1.2,2.545516777038574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,1,power_law_1.2,1.5216896057128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,1,balanced,0.031888000667095184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,1,power_law_1.2,2.619852828979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,1,balanced,0.0662613312403361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,1,power_law_1.2,2.8332096099853517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,1,power_law_1.2,11.25098876953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,1,power_law_1.2,2.8410112380981447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,1,power_law_1.01,15.702522277832031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,1,power_law_1.2,2.904102325439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,1,power_law_1.01,0.02625280022621155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,1,balanced,0.08473599950472514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,1,balanced,0.12914133071899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,1,balanced,0.1314826707045237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,1,power_law_1.2,15.64307861328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,1,power_law_1.01,20.128755187988283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,1,power_law_1.01,0.024985599517822265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,1,power_law_1.01,0.03255040049552917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,1,power_law_1.01,0.04901759922504425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,1,power_law_1.01,0.07196800112724304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,1,balanced,0.0395359992980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,1,balanced,0.05435733497142792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,1,power_law_1.01,0.0849407970905304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,1,power_law_1.01,0.11472640037536622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,1,power_law_1.01,0.04159359931945801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,1,power_law_1.01,0.1355712056159973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,1,power_law_1.2,1.8344512939453126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,1,power_law_1.01,0.14164479970932006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,1,power_law_1.01,0.1355520009994507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,1,power_law_1.01,0.13683199882507324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,1,power_law_1.2,1.965830421447754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,1,power_law_1.2,2.163871955871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,1,balanced,0.11956800023714702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,1,balanced,0.15102932850519815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,1,balanced,0.15292800466219583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,1,power_law_1.2,19.950778198242187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,1,power_law_1.2,2.325312042236328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,1,power_law_1.01,0.1404863953590393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,1,power_law_1.01,0.14172799587249757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,1,power_law_1.01,0.14343039989471434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,1,power_law_1.01,0.14633599519729615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,1,power_law_1.2,3.369676971435547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,1,power_law_1.2,2.849510383605957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,1,power_law_1.01,0.1519744038581848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,1,power_law_1.01,0.15694719552993774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,1,power_law_1.01,0.17045120000839234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,1,power_law_1.01,0.20781440734863282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,1,power_law_1.01,0.23487999439239501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,1,power_law_1.2,3.191276741027832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,1,power_law_1.01,0.2749119997024536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,1,power_law_1.2,3.7102081298828127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,1,power_law_1.01,0.3175168037414551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,1,power_law_1.01,0.06535680294036865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,1,power_law_1.01,0.4065728187561035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,1,power_law_1.01,0.49607038497924805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,1,power_law_1.2,4.068646240234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,1,power_law_1.01,0.672544002532959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,1,power_law_1.2,4.033209609985351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,1,power_law_1.01,0.8573375701904297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,1,power_law_1.01,1.2200127601623536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,1,power_law_1.2,4.336223983764649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,1,power_law_1.2,4.92474250793457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,1,power_law_1.01,1.5969599723815917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,1,balanced,0.2041013240814209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,1,balanced,0.20602132876714072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,1,balanced,0.09203733007113139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,1,power_law_1.2,5.049580764770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,1,power_law_1.2,6.699014282226562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,1,power_law_1.2,6.2441150665283205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,1,power_law_1.01,0.025459200143814087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,1,power_law_1.01,0.0975488007068634
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,1,power_law_1.01,0.029760000109672547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,1,balanced,0.14447466532389322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,1,balanced,0.14444266756375632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,1,power_law_1.01,0.0367680013179779
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,1,power_law_1.01,0.047007998824119566
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,1,power_law_1.01,0.05688959956169128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,1,power_law_1.2,8.442118072509766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,1,power_law_1.01,0.07296000123023987
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,1,power_law_1.01,0.0775488018989563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,1,power_law_1.01,0.030393600463867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,1,power_law_1.01,0.07954559922218322
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,1,power_law_1.01,0.08097919821739197
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,1,power_law_1.01,0.0832319974899292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,1,power_law_1.01,0.08892160058021545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,1,power_law_1.01,0.09216639995574952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,1,power_law_1.01,0.0980288028717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,1,power_law_1.01,0.10963200330734253
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,1,power_law_1.01,0.09666560292243957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,1,balanced,0.20686399936676025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,1,balanced,0.20632533232371011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,1,power_law_1.01,0.101363205909729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,1,balanced,0.20966933170954385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,1,power_law_1.01,0.11519360542297363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,1,power_law_1.2,7.5424957275390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,1,power_law_1.01,0.11980160474777221
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,1,power_law_1.01,0.13774720430374146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,1,power_law_1.01,0.15869439840316774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,1,power_law_1.01,0.17546240091323853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,1,power_law_1.01,0.23036160469055175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,1,power_law_1.01,0.276307201385498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,1,power_law_1.01,0.3699327945709229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,1,power_law_1.01,0.46401281356811525
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,1,power_law_1.01,0.6459648132324218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,1,power_law_1.2,8.751590728759766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,1,power_law_1.01,0.12170239686965942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,1,power_law_1.01,0.8583616256713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,1,balanced,0.13939199844996134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,1,power_law_1.2,0.025567999482154845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,1,power_law_1.01,0.17517440319061278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,1,power_law_1.2,0.024857600033283234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,1,power_law_1.2,0.033395200967788696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,1,power_law_1.2,0.04661119878292084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,1,power_law_1.2,0.07061759829521179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,1,power_law_1.2,0.08113920092582702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,1,power_law_1.2,0.11724159717559815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,1,power_law_1.2,0.13855359554290772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,1,power_law_1.2,0.14225280284881592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,1,power_law_1.2,0.13638399839401244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,1,power_law_1.2,0.13857280015945433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,1,power_law_1.2,0.1443071961402893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,1,power_law_1.2,0.1431040048599243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,1,balanced,0.20866133769353232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,1,power_law_1.2,0.1450816035270691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,1,power_law_1.2,0.14736000299453736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,1,power_law_1.2,11.923737335205079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,1,power_law_1.2,0.1523327946662903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,1,power_law_1.2,0.16075520515441893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,1,power_law_1.2,0.1722048044204712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,1,power_law_1.2,0.2171839952468872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,1,power_law_1.2,0.23766400814056396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,1,power_law_1.2,0.2802047967910767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,1,power_law_1.01,0.04008319973945618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,1,power_law_1.01,0.052742397785186766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,1,power_law_1.2,0.3261375904083252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,1,balanced,0.14761066436767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,1,power_law_1.2,0.416864013671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,1,power_law_1.2,0.04149119853973389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,1,power_law_1.2,0.50382080078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,1,power_law_1.01,0.1837504029273987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,1,power_law_1.01,0.19014400243759155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,1,power_law_1.01,0.1928320050239563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,1,power_law_1.2,0.682809591293335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,1,power_law_1.2,15.224018859863282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,1,balanced,0.21012266476949057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,1,balanced,0.21165867646535239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,1,balanced,0.212719996770223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,1,power_law_1.01,0.07114880084991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,1,balanced,0.15057067076365152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,1,balanced,0.14189866185188293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,1,balanced,0.14249066511789957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,1,balanced,0.1456106702486674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,1,balanced,0.14500266313552856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,1,balanced,0.14524799585342407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,1,balanced,0.1486133337020874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,1,balanced,0.14974400401115417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,1,balanced,0.15341333548227945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,1,balanced,0.20616000890731812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,1,balanced,0.20995734135309854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,1,power_law_1.2,0.06064000129699707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,1,power_law_1.01,0.20024960041046141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,1,balanced,0.14750400185585022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,1,balanced,0.1490666667620341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,1,balanced,0.150736004114151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,1,balanced,0.15235732992490134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,1,balanced,0.1560373306274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,1,balanced,0.16448533535003662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,1,balanced,0.21760533253351846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,1,balanced,0.27189334233601886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,1,balanced,0.2761066754659017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,1,power_law_1.2,0.8653759956359863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,1,power_law_1.2,0.02959359884262085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,1,power_law_1.01,0.09296000003814697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,1,power_law_1.01,0.20234239101409912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,1,power_law_1.2,0.08984320163726807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,1,power_law_1.01,0.21496961116790772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,1,power_law_1.01,0.2184448003768921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,1,balanced,0.24294400215148926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,1,balanced,0.25595200061798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,1,balanced,0.2688373327255249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,1,balanced,0.20494933923085532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,1,balanced,0.21649066607157388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,1,balanced,0.24012800057729086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,1,balanced,0.2642986575762431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,1,balanced,0.3903040091196696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,1,balanced,0.4379040002822876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,1,balanced,0.6140640179316202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,1,balanced,0.7918666998545328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,1,power_law_1.2,0.10746239423751831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,1,balanced,1.1536746819814045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,1,power_law_1.01,0.23827199935913085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,1,power_law_1.2,0.17601920366287233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,1,power_law_1.01,0.254419207572937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,1,balanced,1.5142985979715984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,1,power_law_1.01,0.2862720012664795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,1,power_law_1.2,0.040140798687934874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,1,power_law_1.2,0.051020801067352295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,1,balanced,0.25278933842976886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,1,balanced,0.2714719971021016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,1,power_law_1.01,0.25516159534454347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,1,power_law_1.01,0.2805504083633423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,1,power_law_1.2,0.024134400486946105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,1,power_law_1.01,0.30751359462738037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,1,power_law_1.01,0.12816640138626098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,1,power_law_1.01,0.13303040266036986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,1,power_law_1.01,0.13390719890594482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,1,power_law_1.01,0.13609600067138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,1,power_law_1.01,0.13817600011825562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,1,power_law_1.01,0.13963520526885986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,1,power_law_1.01,0.1541440010070801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,1,power_law_1.01,0.15891200304031372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,1,power_law_1.2,0.029023998975753786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,1,power_law_1.01,0.17566720247268677
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,1,power_law_1.2,0.03521279990673065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,1,power_law_1.01,0.19419519901275634
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,1,power_law_1.2,0.04413439929485321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,1,power_law_1.01,0.21653120517730712
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,1,power_law_1.2,0.05570560097694397
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,1,power_law_1.2,0.07313920259475708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,1,power_law_1.2,0.07740160226821899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,1,power_law_1.2,0.0781823992729187
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,1,power_law_1.2,0.08228480219841003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,1,power_law_1.2,0.186188805103302
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,1,power_law_1.2,0.08359040021896362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,1,power_law_1.2,0.18838399648666382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,1,power_law_1.2,0.0876416027545929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,1,power_law_1.2,0.19753600358963014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,1,power_law_1.2,0.09190400242805481
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,1,power_law_1.2,0.09783040285110474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,1,power_law_1.2,0.1082111954689026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,1,power_law_1.2,0.09599360227584838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,1,power_law_1.2,0.10364799499511719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,1,power_law_1.2,0.11500799655914307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,1,power_law_1.2,0.1245311975479126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,1,power_law_1.2,0.139520001411438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,1,power_law_1.2,0.1508031964302063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,1,power_law_1.2,0.17711999416351318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,1,power_law_1.2,0.23503360748291016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,1,power_law_1.2,0.28088319301605225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,1,power_law_1.2,0.06949759721755981
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,1,power_law_1.2,0.3801599979400635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,1,balanced,0.32381866375605267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,1,balanced,0.3509013255437215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,1,power_law_1.2,0.20152320861816406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,1,power_law_1.01,0.35452799797058104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,1,power_law_1.01,0.40842881202697756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,1,balanced,0.28972800572713214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,1,balanced,0.45053335030873615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,1,balanced,0.48468267917633057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,1,balanced,0.6786293188730875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,1,balanced,0.886303981145223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,1,balanced,1.2717386881510417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,1,balanced,1.6978453000386555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,1,power_law_1.2,0.21568639278411866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,1,power_law_1.2,0.21823360919952392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,1,power_law_1.2,0.09039360284805298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,1,power_law_1.2,0.2191999912261963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,1,power_law_1.2,0.47211518287658694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,1,balanced,2.439136028289795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,1,power_law_1.01,0.281062388420105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,1,power_law_1.01,0.3233855962753296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,1,balanced,3.284458796183268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,1,balanced,0.07237866520881653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,1,balanced,0.11688533425331116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,1,balanced,0.2035520076751709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,1,balanced,0.3856053352355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,1,balanced,0.7152799765268961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,1,balanced,0.7196053663889567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,1,balanced,0.7607519626617432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,1,balanced,0.7624373435974121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,1,balanced,0.8141600290934244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,1,power_law_1.2,1.2329216003417969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,1,balanced,0.8156586488087972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,1,balanced,0.7840586503346761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,1,balanced,0.7832053502400717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,1,balanced,0.7873760064442953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,1,balanced,0.7910772959391276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,1,balanced,0.7971039613087972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,1,balanced,0.8043200174967448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,1,balanced,0.816869338353475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,1,power_law_1.2,0.2422015905380249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,1,balanced,0.9656053384145101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,1,power_law_1.2,0.25724799633026124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,1,balanced,0.9880959987640381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,1,power_law_1.2,0.2880255937576294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,1,power_law_1.2,0.26026880741119385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,1,balanced,1.0467039744059246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,1,power_law_1.2,0.2852479934692383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,1,balanced,1.109610637029012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,1,power_law_1.2,0.3178879976272583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,1,balanced,1.6819732983907063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,1,balanced,1.8189279238382976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,1,balanced,2.6929601033528647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,1,balanced,3.391162554423014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,1,power_law_1.2,0.1261247992515564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,1,balanced,0.4564266602198283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,1,power_law_1.2,0.13097599744796753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,1,balanced,0.5006186564763387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,1,power_law_1.2,0.13419519662857055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,1,balanced,0.7854773203531901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,1,power_law_1.2,0.1366976022720337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,1,balanced,0.8778026898701986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,1,power_law_1.2,0.13636480569839476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,1,balanced,5.1238400141398115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,1,balanced,1.2581493059794109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,1,balanced,1.6337599754333496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,1,power_law_1.2,0.653766393661499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,1,balanced,6.664880116780599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,1,power_law_1.2,0.14677759408950805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,1,power_law_1.2,0.15443840026855468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,1,power_law_1.2,0.16106239557266236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,1,power_law_1.2,0.17802239656448365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,1,power_law_1.01,0.5528512001037598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,1,power_law_1.01,0.6500288009643554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,1,power_law_1.01,0.8352767944335937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,1,power_law_1.01,1.0273664474487305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,1,power_law_1.01,1.4214207649230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,1,power_law_1.01,1.8096511840820313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,1,power_law_1.2,0.19646079540252687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,1,power_law_1.2,0.21799039840698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,1,balanced,0.05065066615740458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,1,power_law_1.01,0.3700031995773315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,1,power_law_1.01,0.4637119770050049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,1,power_law_1.2,0.3695616006851196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,1,power_law_1.01,0.5630591869354248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,1,power_law_1.2,0.42167038917541505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,1,power_law_1.01,0.7562367916107178
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,1,balanced,0.033589333295822144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,1,balanced,0.05403733253479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,1,balanced,0.07819733520348866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,1,balanced,0.12353066603342693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,1,power_law_1.01,0.9394687652587891
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,1,balanced,0.21717333793640137
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,1,balanced,0.22171199321746826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,1,balanced,0.2241493264834086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,1,balanced,0.22750399510065714
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,1,balanced,0.23014400402704874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,1,power_law_1.01,1.3072704315185546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,1,balanced,0.2327786684036255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,1,balanced,0.23724265893300375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,1,balanced,0.256117324034373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,1,balanced,0.25913067658742267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,1,balanced,0.26874132951100665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,1,power_law_1.01,1.6835903167724608
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,1,balanced,0.2655893365542094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,1,balanced,0.26843200127283734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,1,balanced,0.28122133016586304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,1,balanced,0.3026346762975057
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,1,balanced,0.3183786670366923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,1,balanced,0.4390133221944173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,1,balanced,0.06172800064086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,1,power_law_1.01,2.432262420654297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,1,balanced,0.37964268525441486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,1,balanced,0.9603359699249268
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,1,balanced,0.5178933143615723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,1,balanced,0.7410399913787842
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,1,balanced,0.9548426469167074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,1,power_law_1.01,3.167667198181152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,1,balanced,1.3851253191630046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,1,balanced,0.07549333572387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,1,balanced,0.1320853332678477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,1,balanced,1.8887200355529785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,1,power_law_1.01,0.05185279846191406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,1,power_law_1.01,0.07505279779434204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,1,power_law_1.01,0.1292672038078308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,1,power_law_1.01,0.18182400465011597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,1,power_law_1.01,0.2226560115814209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,1,power_law_1.01,0.35047039985656736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,1,power_law_1.01,0.42424960136413575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,1,power_law_1.01,0.4316864013671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,1,power_law_1.01,0.403718376159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,1,power_law_1.01,0.41893758773803713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,1,power_law_1.01,0.40502400398254396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,1,power_law_1.01,0.4057024002075195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,1,power_law_1.01,0.413702392578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,1,power_law_1.01,0.41938557624816897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,1,power_law_1.01,0.4351039886474609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,1,power_law_1.01,0.44716801643371584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,1,power_law_1.01,0.4671040058135986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,1,power_law_1.01,0.5825920104980469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,1,power_law_1.01,0.633951997756958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,1,power_law_1.01,0.06948480010032654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,1,power_law_1.2,0.9069631576538086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,1,power_law_1.01,0.741542387008667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,1,power_law_1.01,0.8409536361694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,1,power_law_1.01,1.0696576118469239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,1,balanced,0.08126399914423625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,1,balanced,0.14457066853841147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,1,power_law_1.01,1.2851903915405274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,1,balanced,0.23477866252263388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,1,balanced,0.4223946730295817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,1,power_law_1.01,1.7433727264404297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,1,power_law_1.01,2.232703971862793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,1,power_law_1.2,0.2819648027420044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,1,balanced,0.22285334269205728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,1,power_law_1.2,0.32604160308837893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,1,balanced,0.3990986744562785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,1,balanced,0.40385599931081134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,1,power_law_1.01,3.1114175796508787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,1,power_law_1.01,4.250035095214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,1,power_law_1.01,0.05978239774703979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,1,power_law_1.01,0.11717120409011841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,1,power_law_1.01,0.2060159921646118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,1,power_law_1.2,0.5754303932189941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,1,power_law_1.2,0.6663296222686768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,1,power_law_1.01,0.03319680094718933
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,1,power_law_1.01,0.05283839702606201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,1,power_law_1.2,0.846399974822998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,1,power_law_1.01,0.07519999742507935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,1,power_law_1.01,0.0772159993648529
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,1,power_law_1.01,0.10300159454345703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,1,power_law_1.01,0.14016640186309814
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,1,power_law_1.01,0.12840319871902467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,1,balanced,0.4339413245519002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,1,power_law_1.01,0.18977279663085939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,1,balanced,0.4360640048980713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,1,power_law_1.01,0.2002432107925415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,1,balanced,0.4363093376159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,1,balanced,0.46797335147857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,1,power_law_1.2,1.6067903518676758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,1,balanced,0.46855998039245605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,1,power_law_1.2,0.07175040245056152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,1,power_law_1.2,0.11930240392684936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,1,power_law_1.2,0.2036288022994995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,1,power_law_1.2,0.26956160068511964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,1,power_law_1.2,0.37424640655517577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,1,power_law_1.2,0.6074431896209717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,1,power_law_1.01,0.20881919860839843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,1,power_law_1.2,0.6860479831695556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,1,power_law_1.2,0.6941760063171387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,1,power_law_1.2,0.758784008026123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,1,power_law_1.2,1.0457471847534179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,1,power_law_1.2,0.7597311973571778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,1,power_law_1.2,0.7471936225891114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,1,power_law_1.01,0.19376640319824218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,1,power_law_1.01,0.21647999286651612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,1,power_law_1.2,0.7709184169769288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,1,power_law_1.01,0.235532808303833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,1,power_law_1.01,0.21552000045776368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,1,power_law_1.2,0.7677184104919433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,1,power_law_1.01,0.23009281158447265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,1,power_law_1.01,0.24826879501342775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,1,power_law_1.2,0.793555212020874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,1,power_law_1.01,0.2644351959228516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,1,power_law_1.2,0.808563232421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,1,power_law_1.01,0.2984832048416138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,1,power_law_1.01,0.2616512060165405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,1,power_law_1.2,0.8492799758911133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,1,power_law_1.01,0.2822720050811768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,1,power_law_1.01,0.30894720554351807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,1,power_law_1.01,0.3191999912261963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,1,power_law_1.2,0.8705216407775879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,1,power_law_1.01,0.40358400344848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,1,power_law_1.01,0.31787519454956054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,1,power_law_1.2,1.0495807647705078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,1,power_law_1.01,0.6081408023834228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,1,power_law_1.01,0.37389440536499025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,1,power_law_1.01,0.38329598903656004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,1,power_law_1.2,1.1277440071105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,1,power_law_1.01,0.44182400703430175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,1,power_law_1.01,0.5139584064483642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,1,power_law_1.2,1.2886015892028808
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,1,power_law_1.01,0.6199295997619629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,1,power_law_1.2,1.460927963256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,1,power_law_1.01,0.8393407821655273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,1,power_law_1.2,0.37873919010162355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,1,power_law_1.2,1.8189823150634765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,1,power_law_1.2,0.4688704013824463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,1,power_law_1.2,0.5627071857452393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,1,balanced,0.4440053304036458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,1,balanced,0.44447465737660724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,1,power_law_1.01,0.36250240802764894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,1,power_law_1.01,0.3849087953567505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,1,power_law_1.01,0.38248960971832274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,1,power_law_1.01,1.050271987915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,1,power_law_1.01,1.4696319580078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,1,power_law_1.2,0.7644608020782471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,1,power_law_1.01,0.6700736045837402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,1,power_law_1.01,2.2787263870239256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,1,power_law_1.01,0.6925568103790283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,1,power_law_1.2,0.04961279928684235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,1,power_law_1.2,2.1457984924316404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,1,power_law_1.2,0.07440639734268188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,1,power_law_1.2,0.11912319660186768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,1,power_law_1.2,0.15877759456634521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,1,power_law_1.2,0.9352512359619141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,1,power_law_1.2,0.20466558933258056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,1,balanced,0.6998453140258789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,1,balanced,0.7014453411102295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,1,balanced,0.7081813017527262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,1,balanced,0.7149706681569418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,1,balanced,0.7201706568400065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,1,balanced,0.7276000181833903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,1,balanced,0.7386240164438883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,1,balanced,0.7467199961344401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,1,balanced,0.7652693589528402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,1,balanced,0.7995839913686117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,1,power_law_1.2,0.34949119091033937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,1,balanced,0.8371040026346842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,1,power_law_1.2,0.4172671794891357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,1,power_law_1.2,0.42838401794433595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,1,balanced,1.387605349222819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,1,power_law_1.2,0.4202752113342285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,1,balanced,1.4574400583902996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,1,power_law_1.2,0.4193920135498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,1,power_law_1.2,0.40955519676208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,1,balanced,2.1193226178487143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,1,power_law_1.2,0.4106112003326416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,1,power_law_1.01,0.6400832176208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,1,power_law_1.2,0.4109824180603027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,1,power_law_1.2,2.8447423934936524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,1,balanced,2.755626678466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,1,power_law_1.2,0.4397568225860596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,1,power_law_1.01,0.6484864234924317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,1,power_law_1.2,0.44614400863647463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,1,power_law_1.01,0.6755648136138916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,1,power_law_1.2,0.4527743816375732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,1,power_law_1.01,0.6761280059814453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,1,power_law_1.2,0.48732800483703614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,1,balanced,4.026538530985515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,1,power_law_1.01,0.6949696063995361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,1,power_law_1.2,0.6015103816986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,1,power_law_1.2,1.309267234802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,1,balanced,0.4185119867324829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,1,power_law_1.2,1.4339200019836427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,1,power_law_1.2,0.6545983791351319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,1,power_law_1.2,0.7606592178344727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,1,power_law_1.01,0.7511360168457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,1,balanced,5.30128542582194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,1,power_law_1.2,0.8659775733947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,1,power_law_1.01,0.7562943935394287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,1,power_law_1.01,0.7403391838073731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,1,power_law_1.2,1.0859904289245605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,1,power_law_1.01,0.7571263790130616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,1,balanced,7.766991933186849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,1,power_law_1.2,1.3085503578186035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,1,power_law_1.01,0.7935935974121093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,1,power_law_1.01,0.8688703536987304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,1,power_law_1.2,1.7661439895629882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,1,power_law_1.01,1.0196991920471192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,1,power_law_1.01,1.178828811645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,1,power_law_1.01,1.4818559646606446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,1,balanced,10.28542391459147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,1,power_law_1.01,1.7911615371704102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,1,power_law_1.01,2.4252288818359373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,1,power_law_1.2,0.06004480123519897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,1,power_law_1.2,0.07832959890365601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,1,power_law_1.2,0.1318400025367737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,1,power_law_1.2,0.17126400470733644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,1,power_law_1.2,0.2297215938568115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,1,power_law_1.01,3.0230207443237305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,1,power_law_1.2,0.3721343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,1,power_law_1.2,0.38431360721588137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,1,power_law_1.2,0.38830718994140623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,1,power_law_1.2,0.6554815769195557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,1,power_law_1.01,4.226252746582031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,1,power_law_1.2,0.661516809463501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,1,power_law_1.2,0.6615488052368164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,1,balanced,0.42267199357350665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,1,balanced,0.4250719944636027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,1,balanced,0.4331680138905843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,1,power_law_1.2,0.6948224067687988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,1,balanced,0.43828264872233075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,1,balanced,0.4456106821695964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,1,power_law_1.2,0.7305215835571289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,1,balanced,0.4599039951960246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,1,power_law_1.2,0.7483520030975341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,1,power_law_1.01,5.370751953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,1,power_law_1.2,0.7892096042633057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,1,power_law_1.2,0.8273664474487304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,1,power_law_1.01,0.7617728233337402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,1,power_law_1.2,0.9058176040649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,1,power_law_1.2,1.049177646636963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,1,power_law_1.2,1.1996479988098145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,1,power_law_1.01,7.644300842285157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,1,power_law_1.2,1.497100830078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,1,power_law_1.2,2.2033920288085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,1,power_law_1.2,1.8197759628295898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,1,power_law_1.2,2.4285760879516602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,1,power_law_1.2,1.6852800369262695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,1,power_law_1.01,9.912281799316407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,1,power_law_1.2,3.0086015701293944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,1,power_law_1.2,3.520671844482422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,1,balanced,0.5546666781107584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,1,power_law_1.01,0.7573503971099853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,1,balanced,0.5790506601333618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,1,power_law_1.2,0.03048959970474243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,1,balanced,0.6233706474304199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,1,power_law_1.01,0.7602367877960206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,1,power_law_1.2,0.05200639963150024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,1,balanced,0.6753760178883871
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,1,power_law_1.2,0.06959360241889953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,1,power_law_1.01,0.7977471828460694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,1,balanced,1.000325361887614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,1,power_law_1.2,0.0920960009098053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,1,power_law_1.01,0.7985663890838623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,1,balanced,1.1002826690673828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,1,power_law_1.2,0.11874560117721558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,1,power_law_1.2,4.965887832641601
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,1,power_law_1.2,0.19100799560546874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,1,power_law_1.01,0.8050944328308105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,1,balanced,1.6050772666931152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,1,power_law_1.2,0.20302720069885255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,1,power_law_1.01,0.8454143524169921
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,1,power_law_1.2,0.21246719360351562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,1,balanced,2.0261866251627603
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,1,power_law_1.2,0.21910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,1,power_law_1.2,0.22554240226745606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,1,power_law_1.2,0.23716480731964112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,1,balanced,2.9534667332967124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,1,power_law_1.2,0.25218560695648196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,1,power_law_1.2,0.26859519481658933
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,1,power_law_1.2,0.30512640476226804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,1,power_law_1.2,0.2715519905090332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,1,balanced,3.948362668355306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,1,power_law_1.2,0.28695681095123293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,1,power_law_1.2,0.33036160469055176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,1,power_law_1.2,0.3320512056350708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,1,power_law_1.2,0.3837440013885498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,1,power_law_1.2,0.39144320487976075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,1,power_law_1.2,0.4525951862335205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,1,power_law_1.2,2.4288639068603515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,1,power_law_1.2,0.5356095790863037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,1,power_law_1.2,0.6597631931304931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,1,power_law_1.2,4.177856063842773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,1,power_law_1.2,0.8566720008850097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,1,balanced,0.14017066359519958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,1,power_law_1.01,1.0167296409606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,1,power_law_1.01,1.0989376068115235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,1,power_law_1.01,1.2547136306762696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,1,power_law_1.01,1.4150976181030273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,1,power_law_1.01,1.7620735168457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,1,power_law_1.01,2.1262527465820313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,1,power_law_1.2,1.07128963470459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,1,power_law_1.01,2.862726402282715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,1,power_law_1.01,3.651123046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,1,power_law_1.2,5.322982406616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,1,balanced,0.24213866392771402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,1,balanced,0.4490026632944743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,1,power_law_1.01,5.154355239868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,1,power_law_1.01,6.972255706787109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,1,balanced,0.8745600382486979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,1,balanced,1.703781286875407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,1,balanced,2.125391960144043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,1,balanced,0.08719467123349507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,1,power_law_1.2,1.4936960220336915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,1,balanced,2.2011839548746743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,1,balanced,2.2065919240315757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,1,balanced,0.1597760021686554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,1,balanced,0.2704213261604309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,1,power_law_1.2,1.8211200714111329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,1,power_law_1.2,6.638406372070312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,1,balanced,2.455952008565267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,1,balanced,2.4594613711039224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,1,power_law_1.2,7.587372589111328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,1,balanced,0.509114662806193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,1,balanced,0.9709546566009521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,1,balanced,1.2112747033437092
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,1,balanced,0.062133332093556724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,1,balanced,0.09130133191744487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,1,balanced,0.14722667137781778
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,1,balanced,0.2629599968592326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,1,balanced,0.49127999941507977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,1,balanced,0.6111520131429037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,1,balanced,0.6149813334147135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,1,power_law_1.2,3.153817558288574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,1,balanced,0.620906670888265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,1,balanced,0.6254026492436727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,1,balanced,0.6291786829630533
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,1,balanced,0.63591468334198
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,1,balanced,0.6436906655629476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,1,balanced,0.6977759997049967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,1,balanced,0.7113866806030273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,1,balanced,0.7263627052307129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,1,balanced,0.7193866570790609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,1,power_law_1.2,3.1677759170532225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,1,balanced,0.7352213064829508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,1,balanced,0.7982347011566162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,1,power_law_1.01,0.13694080114364623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,1,balanced,0.8257919947306315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,1,power_law_1.01,0.23969919681549073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,1,balanced,1.3135093053181965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,1,power_law_1.01,0.4474944114685059
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,1,balanced,1.0405759811401367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,1,power_law_1.01,0.7297088146209717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,1,balanced,1.357375939687093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,1,balanced,3.242512067159017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,1,balanced,1.3601493835449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,1,power_law_1.01,0.9941951751708984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,1,balanced,1.9852800369262695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,1,balanced,2.325530687967936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,1,power_law_1.01,1.6649663925170899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,1,balanced,2.8607680002848306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,1,balanced,2.3737173080444336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,1,power_law_1.01,1.8098432540893554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,1,balanced,2.422229290008545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,1,balanced,3.713695844014486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,1,power_law_1.01,1.9152959823608398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,1,balanced,2.4671786626180015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,1,balanced,3.1578292846679688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,1,power_law_1.01,2.150899124145508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,1,balanced,2.462752024332682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,1,power_law_1.01,2.2183296203613283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,1,balanced,2.4707412719726562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,1,balanced,5.297669410705566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,1,balanced,2.5254294077555337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,1,power_law_1.2,4.0254974365234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,1,power_law_1.01,2.1248640060424804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,1,power_law_1.01,0.08691200017929077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,1,power_law_1.01,2.140889549255371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,1,power_law_1.01,0.1606528043746948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,1,power_law_1.01,0.2695103883743286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,1,power_law_1.01,0.4276864051818848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,1,power_law_1.01,2.154310417175293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,1,power_law_1.01,0.5800064086914063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,1,power_law_1.01,0.9445119857788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,1,power_law_1.01,2.2890111923217775
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,1,power_law_1.2,2.4023296356201174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,1,power_law_1.01,1.1377152442932128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,1,power_law_1.01,2.282431983947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,1,power_law_1.01,1.189907169342041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,1,power_law_1.01,1.152665615081787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,1,power_law_1.01,2.459059143066406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,1,power_law_1.01,1.152940845489502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,1,power_law_1.01,1.0952832221984863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,1,power_law_1.01,2.548966407775879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,1,power_law_1.01,1.1223487854003906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,1,balanced,1.2884746392567952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,1,balanced,1.2898666858673096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,1,power_law_1.01,1.11976957321167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,1,power_law_1.01,3.2192638397216795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,1,power_law_1.01,1.1681792259216308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,1,power_law_1.01,1.1879551887512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,1,power_law_1.01,3.2527870178222655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,1,power_law_1.01,1.2081472396850585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,1,power_law_1.01,1.2805184364318847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,1,power_law_1.01,3.68524169921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,1,power_law_1.01,1.5431872367858888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,1,power_law_1.01,1.6551679611206054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,1,power_law_1.01,4.103660964965821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,1,power_law_1.01,0.0587007999420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,1,balanced,2.771045366923014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,1,power_law_1.01,1.8672128677368165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,1,power_law_1.01,0.08644480109214783
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,1,power_law_1.01,0.14540159702301025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,1,balanced,2.9154720306396484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,1,power_law_1.01,0.2241408109664917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,1,power_law_1.01,2.095392036437988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,1,power_law_1.01,0.2973184108734131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,1,power_law_1.01,5.0272575378417965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,1,power_law_1.01,0.48682241439819335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,1,balanced,3.0281759897867837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,1,power_law_1.01,0.5152383804321289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,1,power_law_1.01,2.5845951080322265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,1,power_law_1.01,0.5390079975128174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,1,balanced,3.2320852279663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,1,power_law_1.01,0.5516032218933106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,1,power_law_1.01,0.09810559749603272
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,1,power_law_1.01,0.5606592178344727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,1,power_law_1.01,3.0909183502197264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,1,power_law_1.01,5.831955337524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,1,power_law_1.01,0.5957759857177735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,1,power_law_1.01,0.6323455810546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,1,balanced,5.505770365397136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,1,power_law_1.01,0.6865280151367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,1,power_law_1.01,0.7707647800445556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,1,power_law_1.01,4.026995086669922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,1,power_law_1.01,0.8491711616516113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,1,balanced,5.1774241129557295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,1,power_law_1.01,0.7316224098205566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,1,power_law_1.01,8.023436737060546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,1,power_law_1.01,0.8202816009521484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,1,power_law_1.01,0.8067456245422363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,1,power_law_1.01,5.174399948120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,1,power_law_1.01,0.9278528213500976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,1,balanced,7.641466776529948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,1,power_law_1.01,0.9612031936645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,1,power_law_1.01,1.1012096405029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,1,balanced,1.2030666669209797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,1,power_law_1.01,1.2620479583740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,1,power_law_1.01,7.080166625976562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,1,power_law_1.01,9.78765411376953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,1,power_law_1.01,1.5133119583129884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,1,balanced,10.370416005452475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,1,power_law_1.01,1.9429439544677733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,1,power_law_1.01,2.4227712631225584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,1,power_law_1.01,9.335711669921874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,1,power_law_1.01,3.338016128540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,1,power_law_1.01,0.18069119453430177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,1,power_law_1.2,0.1463104009628296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,1,balanced,13.042043050130209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,1,power_law_1.2,0.24858241081237792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,1,power_law_1.01,0.29495038986206057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,1,power_law_1.01,13.348396301269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,1,power_law_1.2,0.4593855857849121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,1,power_law_1.2,0.6566847801208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,1,power_law_1.2,0.9266176223754883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,1,power_law_1.01,5.336076736450195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,1,power_law_1.2,1.6519872665405273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,1,power_law_1.2,0.08628479838371277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,1,power_law_1.2,0.16024320125579833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,1,power_law_1.2,1.855615997314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,1,power_law_1.2,0.2688256025314331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,1,power_law_1.2,0.4032896041870117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,1,power_law_1.2,1.9208255767822267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,1,power_law_1.2,0.5220863819122314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,1,power_law_1.2,0.9575872421264648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,1,balanced,17.32000478108724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,1,power_law_1.2,2.1690240859985352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,1,power_law_1.2,1.131929588317871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,1,balanced,1.217471996943156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,1,power_law_1.2,2.2633407592773436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,1,power_law_1.2,1.1893183708190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,1,balanced,1.2203199863433838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,1,power_law_1.01,17.110304260253905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,1,balanced,1.2304906845092773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,1,power_law_1.2,1.14967041015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,1,power_law_1.2,2.1846208572387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,1,balanced,1.2540693283081055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,1,power_law_1.2,1.1508416175842284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,1,balanced,1.2739733060201008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,1,balanced,1.2956639925638835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,1,power_law_1.2,1.1138303756713868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,1,power_law_1.2,2.2248191833496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,1,power_law_1.2,1.1455552101135253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,1,power_law_1.2,2.1879552841186523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,1,power_law_1.2,1.145529556274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,1,power_law_1.2,1.1862848281860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,1,power_law_1.2,2.298931121826172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,1,power_law_1.2,1.2271103858947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,1,power_law_1.2,2.4369407653808595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,1,power_law_1.2,1.2637439727783204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,1,power_law_1.01,0.4583615779876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,1,power_law_1.2,1.3301376342773437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,1,power_law_1.01,0.5972671985626221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,1,power_law_1.2,2.416454315185547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,1,power_law_1.2,1.6120447158813476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,1,power_law_1.2,9.845132446289062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,1,power_law_1.2,2.4994688034057617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,1,power_law_1.2,1.7009151458740235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,1,balanced,0.14913066228230795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,1,balanced,0.2421226700146993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,1,balanced,0.42607466379801434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,1,power_law_1.2,1.9670015335083009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,1,power_law_1.2,3.08984317779541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,1,balanced,0.8232479890187582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,1,balanced,1.5973119735717773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,1,power_law_1.2,2.184646415710449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,1,balanced,3.148042678833008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,1,power_law_1.2,3.171558380126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,1,power_law_1.2,2.6560768127441405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,1,power_law_1.2,0.05858560204505921
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,1,power_law_1.2,0.08677759766578674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,1,balanced,5.050778706868489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,1,power_law_1.2,3.626822280883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,1,power_law_1.2,0.14092799425125122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,1,power_law_1.2,3.0815616607666017
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,1,power_law_1.2,0.2123136043548584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,1,power_law_1.2,0.2794431924819946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,1,power_law_1.2,0.47907199859619143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,1,balanced,5.254709243774414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,1,power_law_1.2,0.514796781539917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,1,power_law_1.2,4.0083263397216795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,1,power_law_1.2,0.5528192043304443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,1,power_law_1.2,4.0667072296142575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,1,power_law_1.2,0.552396821975708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,1,balanced,1.4981813430786133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,1,balanced,5.502469380696614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,1,power_law_1.2,0.5672704219818115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,1,balanced,1.5327787399291992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,1,power_law_1.2,0.6047616004943848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,1,balanced,1.6093279520670574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,1,power_law_1.2,5.061971282958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,1,power_law_1.2,0.6592383861541748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,1,balanced,1.7134613990783691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,1,power_law_1.2,0.7120895862579346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,1,balanced,5.48748779296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,1,power_law_1.2,0.7892864227294922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,1,balanced,2.765872001647949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,1,power_law_1.2,0.09857280254364013
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,1,power_law_1.2,0.8667519569396973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,1,power_law_1.2,0.7507840156555176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,1,power_law_1.2,5.888083267211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,1,power_law_1.01,0.9798975944519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,1,power_law_1.2,0.8499839782714844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,1,power_law_1.01,1.0471424102783202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,1,power_law_1.2,0.8351807594299316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,1,power_law_1.01,1.0643199920654296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,1,power_law_1.2,5.0885566711425785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,1,power_law_1.2,0.9463168144226074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,1,power_law_1.2,0.9843199729919434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,1,balanced,2.9941174189249673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,1,balanced,5.455285390218099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,1,power_law_1.2,1.1183039665222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,1,balanced,3.9178028106689453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,1,power_law_1.2,1.3160191535949708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,1,balanced,5.198853174845378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,1,power_law_1.2,7.652665710449218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,1,balanced,5.35963757832845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,1,power_law_1.2,7.332908630371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,1,balanced,5.327333450317383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,1,power_law_1.2,1.5484992027282716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,1,balanced,5.442223866780599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,1,balanced,6.805173238118489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,1,balanced,5.472527821858724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,1,power_law_1.2,9.457421112060548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,1,balanced,9.16644287109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,1,power_law_1.2,9.404914855957031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,1,balanced,5.478410720825195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,1,power_law_1.2,1.974393653869629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,1,power_law_1.2,0.1809280037879944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,1,power_law_1.2,0.29254400730133057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,1,balanced,0.09994666775067647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,1,balanced,0.0846666693687439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,1,balanced,0.1453439990679423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,1,balanced,5.518074671427409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,1,balanced,0.2573759953180949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,1,balanced,0.474021315574646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,1,balanced,0.9277333418528239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,1,balanced,1.8025919596354167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,1,balanced,6.738010406494141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,1,balanced,3.197264035542806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,1,power_law_1.2,2.422809600830078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,1,balanced,3.320906639099121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,1,balanced,6.661098480224609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,1,balanced,2.9809548060099282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,1,balanced,3.0844319661458335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,1,power_law_1.2,13.049261474609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,1,balanced,2.70362122853597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,1,balanced,6.706106821695964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,1,balanced,2.7420053482055664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,1,balanced,2.768303871154785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,1,balanced,6.758053461710612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,1,balanced,2.8005654017130532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,1,balanced,2.853359858194987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,1,balanced,7.040112177530925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,1,balanced,2.867408116658529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,1,power_law_1.01,1.747270393371582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,1,power_law_1.2,0.42046079635620115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,1,balanced,0.1851466695467631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,1,balanced,0.2972106734911601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,1,balanced,0.09524266918500264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,1,balanced,2.902517318725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,1,power_law_1.2,0.5413311958312989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,1,balanced,3.214085261027018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,1,balanced,7.299018859863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,1,balanced,3.3786398569742837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,1,power_law_1.01,1.7885503768920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,1,balanced,3.5414613087972007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,1,power_law_1.01,1.8332672119140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,1,balanced,7.087082544962565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,1,power_law_1.01,1.8792192459106445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,1,balanced,3.573850631713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,1,power_law_1.01,1.9061119079589843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,1,balanced,3.805994669596354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,1,balanced,0.15083733201026917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,1,balanced,0.26870934168497723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,1,balanced,0.4980693260828654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,1,power_law_1.01,1.9262527465820312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,1,balanced,4.007242520650228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,1,balanced,11.718538920084635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,1,power_law_1.01,2.0781312942504884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,1,power_law_1.01,2.1244480133056642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,1,balanced,4.4226986567179365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,1,power_law_1.01,2.275027275085449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,1,balanced,6.59877332051595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,1,power_law_1.01,2.6155967712402344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,1,balanced,12.978912353515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,1,power_law_1.2,3.3879936218261717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,1,power_law_1.01,2.990687942504883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,1,balanced,0.9510986804962158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,1,power_law_1.2,17.082681274414064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,1,balanced,7.631866455078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,1,power_law_1.2,0.9764415740966796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,1,power_law_1.01,3.687590408325195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,1,power_law_1.2,1.040396785736084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,1,power_law_1.2,1.0830719947814942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,1,power_law_1.01,4.390617752075196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,1,balanced,10.09491221110026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,1,balanced,0.5382560094197592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,1,balanced,18.676639556884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,1,balanced,0.9975360234578451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,1,power_law_1.01,0.15898879766464233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,1,power_law_1.01,0.24887681007385254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,1,power_law_1.01,5.842828750610352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,1,power_law_1.01,0.42430720329284666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,1,balanced,1.825285275777181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,1,power_law_1.01,0.8216640472412109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,1,balanced,2.7123680114746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,1,power_law_1.01,1.453215980529785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,1,balanced,2.7521279652913413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,1,power_law_1.01,7.250444793701172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,1,power_law_1.01,2.075257682800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,1,power_law_1.01,2.5707775115966798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,1,balanced,0.05878399809201559
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,1,balanced,0.08548800150553386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,1,power_law_1.01,3.3746814727783203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,1,balanced,0.13985066612561545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,1,balanced,0.2480213244756063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,1,balanced,0.4639893372853597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,1,power_law_1.01,10.032550048828124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,1,balanced,0.8932106494903564
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,1,balanced,1.3300320307413738
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,1,balanced,1.329919974009196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,1,power_law_1.01,3.7697792053222656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,1,balanced,1.3325973351796467
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,1,balanced,1.337450663248698
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,1,balanced,1.3456692695617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,1,power_law_1.01,3.883814239501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,1,balanced,1.3535946210225422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,1,balanced,1.364986737569173
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,1,balanced,1.3804319699605305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,1,balanced,1.3979040781656902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,1,balanced,1.2370933691660564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,1,power_law_1.01,3.9448062896728517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,1,power_law_1.01,12.814247131347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,1,balanced,1.4135200182596843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,1,balanced,1.2458559672037761
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,1,balanced,1.5509653091430664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,1,balanced,1.2503413359324138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,1,balanced,1.6159253120422363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,1,power_law_1.01,4.162783813476563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,1,power_law_1.01,0.08103039860725403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,1,balanced,4.368624051411946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,1,power_law_1.01,4.282867050170898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,1,balanced,4.377530733744304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,1,balanced,1.6001920700073242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,1,balanced,1.6989067395528157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,1,balanced,4.392112096150716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,1,balanced,1.8342132568359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,1,power_law_1.01,4.348128128051758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,1,power_law_1.2,5.2691902160644535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,1,balanced,2.013925393422445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,1,balanced,4.407866795857747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,1,power_law_1.01,18.30164489746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,1,balanced,2.8627732594807944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,1,power_law_1.01,4.466444778442383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,1,balanced,4.420127868652344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,1,balanced,2.430847962697347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,1,power_law_1.2,1.7819135665893555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,1,balanced,4.446576118469238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,1,power_law_1.2,1.798111915588379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,1,balanced,6.021562576293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,1,power_law_1.01,4.57367057800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,1,power_law_1.2,1.8609216690063477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,1,balanced,4.468949317932129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,1,balanced,3.272069295247396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,1,power_law_1.2,1.9316160202026367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,1,power_law_1.01,0.14357119798660278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,1,power_law_1.01,4.639820861816406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,1,balanced,4.492565472920735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,1,power_law_1.01,0.2519999980926514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,1,power_law_1.2,1.9681663513183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,1,balanced,5.931765238444011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,1,power_law_1.01,0.47205119132995604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,1,power_law_1.2,2.0279808044433594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,1,balanced,4.540213267008464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,1,power_law_1.01,5.434611129760742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,1,power_law_1.2,2.1662464141845703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,1,balanced,4.618602752685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,1,power_law_1.01,0.09416319727897644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,1,power_law_1.2,2.2244800567626952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,1,power_law_1.01,0.14979840517044068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,1,power_law_1.01,23.712544250488282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,1,power_law_1.01,0.26331520080566406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,1,balanced,4.687578519185384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,1,power_law_1.01,5.735494232177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,1,power_law_1.2,2.4014720916748047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,1,power_law_1.01,0.4995584011077881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,1,power_law_1.2,0.14350080490112305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,1,power_law_1.01,0.8465984344482422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,1,power_law_1.2,0.23927040100097657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,1,power_law_1.2,2.7170495986938477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,1,balanced,4.803418795267741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,1,power_law_1.2,0.4251071929931641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,1,power_law_1.01,1.2296704292297362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,1,power_law_1.2,0.8234432220458985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,1,power_law_1.01,6.224595260620117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,1,power_law_1.2,3.070636749267578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,1,power_law_1.01,1.395257568359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,1,balanced,4.918453216552734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,1,power_law_1.2,1.3209728240966796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,1,power_law_1.01,1.8360767364501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,1,power_law_1.2,3.800505447387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,1,power_law_1.2,1.8355520248413086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,1,balanced,5.1632693608601885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,1,power_law_1.01,6.723481750488281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,1,power_law_1.01,3.144153594970703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,1,power_law_1.2,2.246899223327637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,1,power_law_1.2,4.504851150512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,1,balanced,1.989712079366048
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,1,power_law_1.01,0.057196801900863646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,1,power_law_1.01,3.150105667114258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,1,balanced,1.9983466466267903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,1,power_law_1.01,0.08359040021896362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,1,balanced,8.756138483683268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,1,power_law_1.2,3.175436782836914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,1,power_law_1.01,0.1388416051864624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,1,balanced,2.0117012659708657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,1,power_law_1.01,0.24717440605163574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,1,balanced,2.023973306020101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,1,power_law_1.01,7.402777862548828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,1,power_law_1.01,3.402579116821289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,1,power_law_1.2,5.886854553222657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,1,power_law_1.01,0.4287744045257568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,1,balanced,2.034719944000244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,1,power_law_1.01,0.8227456092834473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,1,power_law_1.2,3.6898944854736326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,1,power_law_1.01,0.6090559959411621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,1,balanced,2.0550400416056314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,1,balanced,9.188010533650717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,1,power_law_1.01,0.7254079818725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,1,power_law_1.01,3.4078208923339846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,1,balanced,2.075040022532145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,1,power_law_1.01,0.9262144088745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,1,balanced,2.092853387196859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,1,power_law_1.2,7.280435180664062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,1,power_law_1.01,0.9619775772094726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,1,power_law_1.01,3.474470520019531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,1,balanced,2.128490606943766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,1,power_law_1.01,8.493875122070312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,1,power_law_1.01,1.1668095588684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,1,balanced,2.1886293093363443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,1,balanced,2.249274730682373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,1,power_law_1.01,3.6184513092041017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,1,power_law_1.2,3.830400085449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,1,balanced,13.104751586914062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,1,balanced,3.8913065592447915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,1,power_law_1.01,0.9902079582214356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,1,power_law_1.01,3.668038558959961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,1,power_law_1.2,3.8770751953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,1,power_law_1.01,1.0541119575500488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,1,balanced,3.991722742716471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,1,power_law_1.01,10.179590606689453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,1,power_law_1.2,10.063750457763671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,1,power_law_1.01,1.076364803314209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,1,power_law_1.01,3.7741825103759767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,1,power_law_1.01,1.0943679809570312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,1,power_law_1.2,4.039424133300781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,1,balanced,5.778805414835612
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,1,power_law_1.01,1.157222366333008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,1,power_law_1.01,3.8923583984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,1,power_law_1.01,1.2032447814941407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,1,power_law_1.2,4.13359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,1,balanced,17.41376495361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,1,power_law_1.01,1.299449634552002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,1,balanced,7.546384175618489
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,1,power_law_1.01,1.4580863952636718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,1,power_law_1.01,4.293100738525391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,1,power_law_1.01,11.962937927246093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,1,power_law_1.2,4.240550231933594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,1,power_law_1.01,1.822470474243164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,1,power_law_1.2,12.750707244873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,1,power_law_1.01,1.609516716003418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,1,power_law_1.01,4.527219009399414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,1,power_law_1.01,1.5752448081970214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,1,balanced,9.538666407267252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,1,power_law_1.2,4.343360137939453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,1,power_law_1.01,1.9673664093017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,1,power_law_1.01,2.0121984481811523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,1,power_law_1.01,1.8566272735595704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,1,power_law_1.01,5.230233764648437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,1,power_law_1.2,4.4390911102294925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,1,power_law_1.01,2.319705581665039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,1,power_law_1.01,15.684390258789062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,1,balanced,13.029195149739584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,1,power_law_1.01,2.1935680389404295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,1,balanced,25.13072458902995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,1,power_law_1.2,4.5307777404785154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,1,power_law_1.01,5.865248107910157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,1,power_law_1.01,2.8076671600341796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,1,power_law_1.2,0.08082559704780579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,1,power_law_1.2,5.537203216552735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,1,power_law_1.2,0.14531840085983277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,1,power_law_1.01,3.018387222290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,1,power_law_1.2,0.2551104068756104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,1,power_law_1.01,7.211641693115235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,1,power_law_1.2,0.47336320877075194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,1,power_law_1.2,0.7563583850860596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,1,power_law_1.2,6.011430358886718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,1,power_law_1.01,3.9560897827148436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,1,power_law_1.2,1.067296028137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,1,balanced,18.514507293701172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,1,power_law_1.2,1.4322239875793457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,1,power_law_1.01,8.497984313964844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,1,power_law_1.01,19.72321319580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,1,power_law_1.2,1.9496320724487304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,1,power_law_1.2,6.514310455322265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,1,power_law_1.01,6.466361236572266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,1,power_law_1.2,1.9292800903320313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,1,power_law_1.2,1.9764095306396485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,1,power_law_1.01,1.9621311187744142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,1,power_law_1.2,7.044198608398437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,1,power_law_1.2,1.928019142150879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,1,power_law_1.01,11.266944122314452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,1,power_law_1.01,2.073753547668457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,1,power_law_1.2,2.0258623123168946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,1,balanced,23.88451639811198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,1,power_law_1.2,2.0741567611694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,1,power_law_1.2,8.036876678466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,1,power_law_1.2,2.162553596496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,1,power_law_1.2,2.238502311706543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,1,power_law_1.2,2.2289407730102537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,1,power_law_1.2,0.09388800263404846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,1,power_law_1.01,13.953523254394531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,1,power_law_1.2,8.971839904785156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,1,power_law_1.2,0.14932479858398437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,1,power_law_1.2,2.3477439880371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,1,power_law_1.2,0.266214394569397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,1,power_law_1.2,0.5000832080841064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,1,power_law_1.2,0.05691519975662231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,1,power_law_1.2,0.7784639835357666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,1,power_law_1.2,2.9056255340576174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,1,power_law_1.2,1.07456636428833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,1,power_law_1.2,3.092639923095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,1,power_law_1.2,1.2743040084838868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,balanced,0.0553653339544932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,1,power_law_1.2,10.585100555419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,1,power_law_1.2,0.08419839739799499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,1,power_law_1.2,1.7680320739746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,1,power_law_1.2,0.13893760442733766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,1,power_law_1.2,3.3947776794433593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,1,power_law_1.2,0.2473344087600708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,1,power_law_1.2,18.194566345214845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,1,power_law_1.2,0.39064960479736327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,1,power_law_1.2,3.009267234802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,1,power_law_1.2,0.5447999954223632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,1,power_law_1.2,3.654790496826172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,1,power_law_1.01,19.262623596191407
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,1,power_law_1.2,0.6585279941558838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,1,power_law_1.2,3.0824832916259766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,1,power_law_1.2,0.8969920158386231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,1,power_law_1.2,4.265030288696289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,1,power_law_1.2,0.9399744033813476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,1,power_law_1.2,12.451519775390626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,1,power_law_1.2,3.2740734100341795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,1,power_law_1.2,0.9606016159057618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,1,power_law_1.01,1.9729536056518555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,1,power_law_1.2,1.014022445678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,1,power_law_1.2,4.771596908569336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,1,power_law_1.2,3.401830291748047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,1,power_law_1.2,1.0737792015075684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,1,power_law_1.2,1.096448040008545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,1,power_law_1.2,3.4568126678466795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,balanced,0.09091732899347942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,1,power_law_1.2,1.1692864418029785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,balanced,0.16030399998029074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,balanced,0.28520532449086505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,1,power_law_1.2,5.715475082397461
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,1,power_law_1.2,1.2512448310852051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,1,power_law_1.2,3.472652816772461
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,1,power_law_1.2,1.3412799835205078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,1,power_law_1.2,16.237574768066406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,1,power_law_1.2,1.5026623725891113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,1,power_law_1.2,6.892050933837891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,1,power_law_1.2,3.650592041015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,1,power_law_1.01,24.485920715332032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,1,power_law_1.2,1.8648000717163087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,balanced,0.041946664452552795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,balanced,0.06433066725730896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,balanced,0.10355200370152791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,balanced,0.17219199736913046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,balanced,0.3100159962972005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,1,power_law_1.2,1.6833536148071289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,balanced,0.5764853159586588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,balanced,0.6850293477376302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,1,power_law_1.2,3.700851058959961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,balanced,0.6849493185679117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,balanced,0.6297599871953329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,1,power_law_1.2,2.056934356689453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,balanced,0.6319466829299927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,balanced,0.6031306584676107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,balanced,0.6041226784388224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,balanced,0.6063839991887411
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,1,power_law_1.2,1.9415296554565429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,1,power_law_1.2,3.989388656616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,1,power_law_1.2,8.986930847167969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,1,power_law_1.2,2.434726333618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,1,power_law_1.2,4.366924667358399
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,1,power_law_1.2,2.309459114074707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,balanced,0.6138613224029541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,balanced,0.6200106541315714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,balanced,0.6229120095570883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,balanced,0.6346346537272135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,1,power_law_1.2,19.862655639648438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,balanced,0.7826399803161621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,1,power_law_1.01,2.084659194946289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,1,power_law_1.2,2.9125696182250977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,balanced,0.7973173459370931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,1,power_law_1.2,4.706086349487305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,balanced,0.836405356725057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,balanced,0.8775200049082438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,1,power_law_1.2,10.696281433105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,1,power_law_1.01,2.0769792556762696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,balanced,0.9577386379241943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,balanced,1.0463840166727703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,1,power_law_1.2,3.1863040924072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,1,power_law_1.01,2.1681024551391603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,balanced,1.5808107058207195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,1,power_law_1.2,23.555641174316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,1,power_law_1.2,5.521023941040039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,balanced,1.7614186604817708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,power_law_1.01,0.0525056004524231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,balanced,0.5414026578267416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,1,power_law_1.01,2.2229312896728515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,balanced,1.0427199999491374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,balanced,2.570005257924398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,1,power_law_1.2,4.053932952880859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,1,power_law_1.01,2.2646528244018556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,balanced,3.3392747243245444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,1,power_law_1.2,6.203302383422852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,power_law_1.01,0.044512000679969785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,power_law_1.01,0.08923519849777221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,power_law_1.01,0.06349440217018128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,1,power_law_1.01,2.3387327194213867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,power_law_1.01,0.10378880500793457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,power_law_1.01,0.15982719659805297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,power_law_1.01,0.17041280269622802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,power_law_1.01,0.28211839199066163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,balanced,0.031354665756225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,balanced,0.04424533247947693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,balanced,0.06492266555627187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,power_law_1.01,0.24896640777587892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,power_law_1.01,0.41907200813293455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,1,power_law_1.2,6.900358581542969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,power_law_1.01,0.3081792116165161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,power_law_1.01,0.5598080158233643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,power_law_1.01,0.5129280090332031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,balanced,0.05215999980767568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,power_law_1.01,0.8179967880249024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,balanced,0.10028800368309021
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,balanced,0.17063466707865396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,balanced,0.30933332443237305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,balanced,0.31225599845250446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,balanced,0.06880533198515575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,balanced,0.3157973289489746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,balanced,0.11475732922554016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,balanced,0.3169493277867635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,balanced,0.18157867590586343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,power_law_1.01,0.5391295909881592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,balanced,0.31829333305358887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,power_law_1.01,0.8214400291442872
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,balanced,0.3233120044072469
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,balanced,0.3285813331604004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,power_law_1.01,0.5051775932312011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,balanced,0.3311840097109477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,power_law_1.01,0.9275263786315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,power_law_1.01,0.5152703762054444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,power_law_1.01,0.948249626159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,1,power_law_1.2,7.525862121582032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,power_law_1.01,0.9393280029296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,balanced,1.0924373467763264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,power_law_1.01,0.9562175750732422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,balanced,1.0945813655853271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,balanced,0.3389813502629598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,power_law_1.01,0.9838656425476074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,balanced,0.358730673789978
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,balanced,0.3649173180262248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,balanced,0.3818933169047038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,power_law_1.01,1.006220817565918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,balanced,0.3821813265482585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,1,power_law_1.01,2.798681640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,balanced,0.3978240092595418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,balanced,0.43438398838043213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,power_law_1.01,1.0248959541320801
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,balanced,0.46299199263254803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,power_law_1.01,0.05381760001182556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,balanced,0.6357226769129435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,1,power_law_1.2,8.95257568359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,power_law_1.01,1.0442751884460448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,balanced,0.5626346667607626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,1,power_law_1.01,2.8903615951538084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,balanced,1.4242293039957683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,power_law_1.01,1.0786944389343263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,balanced,0.3337920109430949
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,balanced,0.8436053593953451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,power_law_1.01,0.500486421585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,power_law_1.01,1.2803327560424804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,balanced,1.2024586995442708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,power_law_1.01,0.5089983940124512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,1,power_law_1.01,3.1943679809570313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,balanced,1.5535999933878581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,power_law_1.01,1.3478912353515624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,power_law_1.01,0.5263296127319336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,power_law_1.01,0.5424960136413575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,power_law_1.01,1.4875904083251954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,power_law_1.01,0.5545407772064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,1,power_law_1.01,3.510176086425781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,power_law_1.01,0.5696512222290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,power_law_1.01,1.6211456298828124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,power_law_1.01,0.5859776020050049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,power_law_1.01,0.7378176212310791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,power_law_1.01,1.9117504119873048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,1,power_law_1.2,11.494924926757813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,power_law_1.01,0.7690624237060547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,1,power_law_1.01,3.9301185607910156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,power_law_1.01,0.8568127632141114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,power_law_1.01,2.1561151504516602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,power_law_1.01,0.9434176445007324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,power_law_1.01,1.119859218597412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,power_law_1.01,2.806252861022949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,balanced,0.6153653462727865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,1,power_law_1.01,4.467500686645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,balanced,0.6181279818216959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,balanced,0.6198773384094238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,power_law_1.01,0.029363200068473816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,power_law_1.01,1.2995136260986329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,power_law_1.01,0.04568319916725159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,balanced,1.174021323521932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,power_law_1.01,1.671820831298828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,power_law_1.01,3.491436767578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,balanced,1.1759466330210369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,1,power_law_1.01,5.618899154663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,power_law_1.01,2.0223295211791994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,1,power_law_1.2,14.2291259765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,power_law_1.01,4.632243347167969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,power_law_1.01,2.732512092590332
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,power_law_1.01,0.06317440271377564
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,power_law_1.01,0.09844480156898498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,power_law_1.01,0.13848960399627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,power_law_1.01,0.06632959842681885
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,power_law_1.01,0.18238719701766967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,power_law_1.01,0.11349120140075683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,1,power_law_1.01,6.457574462890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,power_law_1.01,3.6514560699462892
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,power_law_1.01,0.2375744104385376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,power_law_1.01,0.179366397857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,power_law_1.2,0.054527997970581055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,power_law_1.01,0.2489919900894165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,power_law_1.01,5.768928146362304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,power_law_1.2,0.09210879802703857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,power_law_1.01,0.2584383964538574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,power_law_1.2,0.16429439783096314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,power_law_1.2,0.05247359871864319
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,power_law_1.01,0.2633919954299927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,power_law_1.2,0.06603519916534424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,power_law_1.2,0.2761087894439697
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,power_law_1.01,0.27241599559783936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,power_law_1.2,0.10733439922332763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,power_law_1.2,0.37934079170227053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,power_law_1.01,0.27601919174194334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,power_law_1.2,0.16374399662017822
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,power_law_1.01,0.2896575927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,power_law_1.2,0.49596800804138186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,power_law_1.2,0.22339200973510742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,power_law_1.01,0.30511999130249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,power_law_1.2,0.7936063766479492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,1,power_law_1.01,8.508230590820313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,power_law_1.2,0.29151999950408936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,power_law_1.01,0.3352895975112915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,power_law_1.2,0.8243519783020019
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,power_law_1.01,0.35961599349975587
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,power_law_1.01,0.41336960792541505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,power_law_1.2,0.9268032073974609
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,power_law_1.01,0.39193599224090575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,power_law_1.2,0.9230912208557129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,power_law_1.01,0.44805121421813965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,1,power_law_1.2,19.386898803710938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,power_law_1.01,0.4574912071228027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,power_law_1.2,0.9458239555358887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,balanced,1.018783966700236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,power_law_1.01,0.5359871864318848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,power_law_1.2,0.9727359771728515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,power_law_1.2,0.4861760139465332
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,power_law_1.01,0.5523071765899659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,power_law_1.2,0.521395206451416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,power_law_1.2,0.9816512107849121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,power_law_1.01,0.6549312114715576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,power_law_1.2,0.4983039855957031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,power_law_1.2,0.9998656272888183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,1,power_law_1.01,10.582201385498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,power_law_1.2,0.5199808120727539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,power_law_1.2,1.0306559562683106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,power_law_1.2,0.49855360984802244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,balanced,1.0224533081054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,power_law_1.2,0.5158592224121094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,balanced,1.030949354171753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,power_law_1.2,0.5237504005432129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,balanced,1.0355359713236492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,power_law_1.2,0.5567039966583252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,balanced,1.0418612957000732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,power_law_1.2,0.5539775848388672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,balanced,1.0554880301157634
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,power_law_1.01,0.8092032432556152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,balanced,1.0641760031382244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,power_law_1.2,0.05367040038108826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,balanced,1.0732213656107585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,power_law_1.2,0.06565759778022766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,balanced,1.1418879826863606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,power_law_1.01,1.0016960144042968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,power_law_1.2,1.0605759620666504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,balanced,1.0905226866404216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,power_law_1.2,0.11490559577941895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,balanced,1.1431360244750977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,balanced,1.125104029973348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,power_law_1.2,0.1721791982650757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,power_law_1.2,1.1169919967651367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,balanced,1.1469919681549072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,power_law_1.01,1.3490943908691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,balanced,1.1578293641408284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,power_law_1.2,0.23800959587097167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,balanced,1.1558240254720051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,power_law_1.01,0.2540607929229736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,power_law_1.2,1.3453120231628417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,balanced,1.2236106395721436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,balanced,1.1685439745585124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,power_law_1.2,0.3102463960647583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,power_law_1.2,0.5834303855895996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,balanced,1.296282688776652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,power_law_1.01,2.1450368881225588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,power_law_1.2,0.4547071933746338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,power_law_1.2,1.4464960098266602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,power_law_1.2,0.6077631950378418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,power_law_1.2,0.4668799877166748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,balanced,2.1469599405924478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,power_law_1.2,0.7559552192687988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,power_law_1.2,1.545907211303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,power_law_1.2,0.8139519691467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,power_law_1.2,0.803321647644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,power_law_1.2,0.8180352210998535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,power_law_1.2,1.6775039672851562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,1,power_law_1.2,24.456166076660157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,power_law_1.2,0.9012672424316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,power_law_1.2,0.8725824356079102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,balanced,1.1732160250345867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,power_law_1.2,0.9747391700744629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,balanced,1.1762879689534504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,power_law_1.2,2.0433536529541017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,power_law_1.2,0.8742015838623047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,power_law_1.2,1.159001636505127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,power_law_1.2,0.9028160095214843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,balanced,2.2723946571350098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,power_law_1.2,2.264473533630371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,power_law_1.2,0.9395263671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,power_law_1.2,1.3386560440063477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,power_law_1.2,0.9801407814025879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,balanced,3.2943038940429688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,power_law_1.2,1.7431743621826172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,power_law_1.2,2.8046016693115234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,power_law_1.2,1.0213695526123048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,power_law_1.2,1.0733823776245117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,power_law_1.2,2.0881792068481446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,balanced,4.299567858378093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,power_law_1.2,1.2389887809753417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,power_law_1.2,3.4430591583251955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,power_law_1.2,1.349619197845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,power_law_1.2,2.8504192352294924
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,power_law_1.2,0.02961919903755188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,power_law_1.2,0.04585599899291992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,balanced,6.2929121653238935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,power_law_1.2,1.599443244934082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,power_law_1.2,0.06378880143165588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,power_law_1.2,4.54890251159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,power_law_1.2,0.09062399864196777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,power_law_1.2,3.6471038818359376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,power_law_1.2,1.8263616561889648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,power_law_1.2,0.12616959810256959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,power_law_1.2,0.16902400255203248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,power_law_1.2,0.2357248067855835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,power_law_1.2,2.3271488189697265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,power_law_1.2,0.25033600330352784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,balanced,8.306079864501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,power_law_1.2,0.25629439353942873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,power_law_1.2,6.139910507202148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,power_law_1.2,0.2581439971923828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,power_law_1.01,0.3495424032211304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,power_law_1.2,2.803980827331543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,power_law_1.2,0.27368319034576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,power_law_1.01,0.464518404006958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,power_law_1.2,0.27959680557250977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,power_law_1.01,0.4794623851776123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,power_law_1.2,0.2890496015548706
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,power_law_1.2,0.3145792007446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,power_law_1.2,3.735347366333008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,power_law_1.2,0.3463360071182251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,power_law_1.2,0.37630720138549806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,power_law_1.2,0.4333888053894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,power_law_1.2,0.4101568222045898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,power_law_1.2,0.4765439987182617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,power_law_1.2,4.632339096069336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,power_law_1.2,0.4776000022888184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,balanced,1.385200023651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,power_law_1.2,0.5609471797943115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,balanced,1.4000693957010906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,power_law_1.2,0.5784768104553223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,balanced,1.441333293914795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,power_law_1.2,0.6744256019592285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,balanced,1.4847680727640789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,balanced,1.5795200665791829
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,power_law_1.2,0.8431808471679687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,power_law_1.2,6.436479949951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,balanced,1.7135732968648274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,power_law_1.2,1.0147071838378907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,balanced,2.746122678120931
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,power_law_1.2,1.3762175559997558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,balanced,2.9309867223103843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,power_law_1.2,2.340255928039551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,balanced,4.028698603312175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,power_law_1.2,8.212818908691407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,balanced,5.106901486714681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,1,balanced,0.047653332352638245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,power_law_1.01,0.8281920433044434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,power_law_1.01,0.8399616241455078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,power_law_1.01,0.8808832168579102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,power_law_1.01,0.8926464080810547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,power_law_1.01,0.8904704093933106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,power_law_1.01,0.9077568054199219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,power_law_1.01,0.9655743598937988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,power_law_1.01,0.9823488235473633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,power_law_1.01,1.0416447639465332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,1,balanced,0.0765173335870107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,power_law_1.01,1.1556735992431642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,power_law_1.01,1.2812159538269043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,power_law_1.01,1.5315008163452148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,power_law_1.01,1.7746240615844726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,power_law_1.01,2.2603008270263674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,1,balanced,0.03643200049797694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,1,balanced,0.05429866909980774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,1,balanced,0.09054399530092876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,power_law_1.01,2.7506431579589843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,1,power_law_1.01,0.03603839874267578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,1,power_law_1.01,0.053600001335144046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,1,power_law_1.01,0.08963840007781983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,power_law_1.01,3.7136001586914062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,1,balanced,0.14081066846847534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,power_law_1.01,4.653113555908203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,1,power_law_1.2,0.047417598962783816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,1,balanced,0.14685333768526712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,1,balanced,0.25030932823816937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,1,balanced,0.3278026580810547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,1,power_law_1.01,0.12944639921188356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,1,power_law_1.01,0.16867200136184693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,1,power_law_1.01,0.2588992118835449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,power_law_1.01,6.473664093017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,1,power_law_1.2,0.03500800132751465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,1,power_law_1.2,0.053548800945281985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,1,power_law_1.2,0.08759679794311523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,power_law_1.01,8.264640045166015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,1,power_law_1.01,0.04663040041923523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,1,power_law_1.01,0.31057279109954833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,1,balanced,0.38537601629892987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,1,power_law_1.01,0.33098878860473635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,1,balanced,0.3853600025177002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,1,power_law_1.2,0.07227519750595093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,1,power_law_1.2,0.12038400173187255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,1,balanced,0.248416006565094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,1,balanced,0.4492586851119995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,1,power_law_1.2,0.15722880363464356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,1,balanced,0.24938666820526123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,1,balanced,0.8517920176188151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,1,power_law_1.2,0.25834879875183103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,1,balanced,0.14964800079663595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,1,balanced,0.2595946590105693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,1,balanced,0.47832000255584717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,1,balanced,0.3630559841791789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,1,power_law_1.01,0.31043200492858886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,1,balanced,0.36593600114186603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,1,power_law_1.01,0.3228287935256958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,1,power_law_1.2,0.3227519989013672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,1,power_law_1.01,0.07355520129203796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,1,power_law_1.2,0.3347584009170532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,1,balanced,0.9178986549377441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,1,balanced,1.7675414085388184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,1,power_law_1.2,0.1342144012451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,1,balanced,2.5946240425109863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,1,balanced,0.4323413372039795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,1,balanced,0.57533331712087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,1,balanced,1.6707733472188313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,1,balanced,0.5783733526865641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,1,balanced,3.257242520650228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,1,balanced,4.794693311055501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,1,balanced,0.3555893500645955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,1,power_law_1.01,0.32451839447021485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,1,power_law_1.2,0.3147775888442993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,1,power_law_1.2,0.3281471967697144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,1,balanced,2.969226519266764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,1,power_law_1.01,0.1387071967124939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,1,balanced,3.0948588053385415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,1,balanced,0.35706134637196857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,1,balanced,4.992282549540202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,1,balanced,0.3572853406270345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,1,balanced,0.3635626633961995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,1,balanced,0.36799466609954834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,1,balanced,0.3731733163197835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,1,power_law_1.01,0.3194688081741333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,1,power_law_1.01,0.3345279932022095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,1,power_law_1.01,0.341215991973877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,1,balanced,5.345343907674153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,1,power_law_1.01,0.3515327930450439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,1,power_law_1.2,0.19211519956588746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,1,power_law_1.01,0.3609344005584717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,1,power_law_1.01,0.376198410987854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,1,power_law_1.2,0.33507840633392333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,1,balanced,0.3781919876734416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,1,balanced,2.7769654591878257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,1,balanced,2.819509188334147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,1,balanced,0.5815146764119467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,1,balanced,0.5830080111821493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,1,balanced,5.394677480061849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,1,power_law_1.01,0.20974080562591552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,1,power_law_1.2,0.3289088010787964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,1,power_law_1.2,0.3415168046951294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,1,power_law_1.01,0.46808958053588867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,1,balanced,5.447813034057617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,1,power_law_1.2,0.35941119194030763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,1,balanced,0.45909865697224933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,1,power_law_1.01,0.49927678108215334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,1,balanced,0.4707466761271159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,1,power_law_1.2,0.35917439460754397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,1,balanced,0.5055306752522787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,1,power_law_1.01,0.5622208118438721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,1,power_law_1.2,0.3688767910003662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,1,balanced,0.5412799914677938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,1,power_law_1.2,0.38265600204467776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,1,balanced,0.8095786571502686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,1,power_law_1.01,0.6283328056335449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,1,balanced,0.8751359780629476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,1,power_law_1.01,0.7729216098785401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,1,power_law_1.2,0.24705920219421387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,1,balanced,1.2148640155792236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,1,power_law_1.2,0.44712319374084475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,1,power_law_1.01,0.9250176429748536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,1,balanced,1.3672854105631511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,1,balanced,0.5848000049591064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,1,power_law_1.2,0.46467838287353513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,1,power_law_1.01,1.2244864463806153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,1,balanced,2.077002684275309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,1,power_law_1.01,1.5302592277526856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,1,balanced,2.626416047414144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,1,balanced,2.728933334350586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,1,power_law_1.01,2.131180763244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,1,power_law_1.01,2.868940734863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,1,power_law_1.2,0.463699197769165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,1,balanced,5.224874814351399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,1,power_law_1.01,0.2886399984359741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,1,power_law_1.2,0.510425615310669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,1,power_law_1.01,0.4520576000213623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,1,power_law_1.2,0.5827072143554688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,1,power_law_1.01,0.47632641792297364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,1,power_law_1.2,0.6528319835662841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,1,power_law_1.2,0.8023167610168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,1,power_law_1.2,0.954918384552002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,1,balanced,2.652682622273763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,1,power_law_1.2,1.2460800170898438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,1,balanced,2.7045278549194336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,1,power_law_1.01,0.2452415943145752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,1,power_law_1.2,1.5470911979675293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,1,balanced,2.7680107752482095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,1,power_law_1.01,0.44915199279785156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,1,balanced,2.8241440455118814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,1,power_law_1.01,0.8552319526672363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,1,balanced,2.8499679565429688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,1,power_law_1.2,0.49085440635681155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,1,balanced,5.228138605753581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,1,power_law_1.2,0.5165376186370849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,1,balanced,2.92851193745931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,1,balanced,5.435173034667969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,1,balanced,0.5872960090637207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,1,balanced,0.5882720152537028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,1,power_law_1.2,2.15731201171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,1,balanced,0.5926773150761923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,1,balanced,0.6002613306045532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,1,balanced,5.449978510538737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,1,balanced,0.6070026556650797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,1,balanced,0.7231787045796713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,1,power_law_1.01,1.4725184440612793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,1,balanced,5.484560012817383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,1,power_law_1.01,2.1271039962768556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,1,power_law_1.2,0.5239871978759766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,1,power_law_1.2,2.763667106628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,1,power_law_1.01,0.4884672164916992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,1,power_law_1.01,0.5013631820678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,1,balanced,5.516133626302083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,1,power_law_1.01,3.557183837890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,1,balanced,3.3179200490315757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,1,balanced,5.767429351806641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,1,balanced,3.535674730936686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,1,balanced,0.1659893294175466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,1,balanced,0.28100266059239704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,1,balanced,3.8067359924316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,1,balanced,0.5006293455759684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,1,power_law_1.01,0.5128640174865723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,1,balanced,4.101722717285156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,1,power_law_1.01,0.14743679761886597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,1,power_law_1.01,0.2551424026489258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,1,power_law_1.01,0.4750016212463379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,1,power_law_1.01,4.030342483520508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,1,power_law_1.01,4.188313674926758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,1,balanced,6.130330403645833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,1,balanced,0.6738826433817545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,1,balanced,6.405354817708333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,1,balanced,0.94705597559611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,1,balanced,1.8257546424865723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,1,power_law_1.2,0.5559487819671631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,1,balanced,6.175834655761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,1,power_law_1.01,0.8178751945495606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,1,power_law_1.2,0.5674496173858643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,1,balanced,6.774602890014648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,1,power_law_1.2,0.5912576198577881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,1,power_law_1.01,1.1615679740905762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,1,power_law_1.01,1.9503360748291017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,1,balanced,9.539621353149414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,1,balanced,7.023248036702474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,1,power_law_1.2,0.6175680160522461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,1,power_law_1.01,4.523712158203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,1,power_law_1.2,0.6461631774902343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,1,power_law_1.2,0.6994751930236817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,1,balanced,7.306789398193359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,1,balanced,10.173866907755533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,1,power_law_1.01,4.631628799438476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,1,power_law_1.01,0.5347519874572754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,1,power_law_1.01,0.5478911876678467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,1,power_law_1.01,0.5817855834960938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,1,power_law_1.01,0.5921599864959717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,1,power_law_1.01,0.6374015808105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,1,power_law_1.01,0.6764671802520752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,1,balanced,12.243802388509115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,1,power_law_1.01,2.3453760147094727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,1,balanced,16.416815439860027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,1,power_law_1.01,2.4247039794921874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,1,balanced,2.693888028462728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,1,balanced,12.44271469116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,1,balanced,2.7113653818766275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,1,balanced,0.672165314356486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,1,balanced,2.7401707967122397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,1,power_law_1.01,4.60552978515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,1,balanced,20.02220280965169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,1,power_law_1.2,0.7648255825042725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,1,balanced,17.734869639078777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,1,power_law_1.01,2.3281343460083006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,1,balanced,0.6970773537953695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,1,power_law_1.01,2.443891143798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,1,power_law_1.01,0.7327744007110596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,1,power_law_1.01,4.710303878784179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,1,balanced,19.264911651611328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,1,power_law_1.01,4.910188674926758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,1,power_law_1.01,4.9048511505126955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,1,balanced,0.7859946886698405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,1,balanced,0.8221866289774576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,1,power_law_1.01,5.078073501586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,1,balanced,1.1830986340840657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,1,balanced,1.2572373549143474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,1,power_law_1.01,2.306438446044922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,1,balanced,1.7492373784383137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,1,balanced,2.0060319900512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,1,power_law_1.01,5.265369415283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,1,balanced,3.1603574752807617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,1,balanced,3.7391573588053384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,1,balanced,30.272201538085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,1,power_law_1.01,0.16343679428100585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,1,power_law_1.2,0.7282879829406739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,1,power_law_1.01,0.2805056095123291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,1,power_law_1.01,0.4996352195739746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,1,power_law_1.01,5.428300857543945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,1,power_law_1.01,2.3723840713500977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,1,power_law_1.01,0.7129087924957276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,1,balanced,4.404357274373372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,1,power_law_1.2,0.2471935987472534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,1,balanced,4.422517458597819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,1,power_law_1.2,0.446073579788208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,1,power_law_1.01,2.4302976608276365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,1,power_law_1.2,0.7825727939605713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,1,power_law_1.2,0.8541695594787597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,1,balanced,4.453680038452148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,1,power_law_1.01,2.4977344512939452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,1,balanced,4.483946800231934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,1,power_law_1.01,2.5543743133544923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,1,power_law_1.01,2.6261119842529297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,1,balanced,4.516090710957845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,1,power_law_1.01,0.7665855884552002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,1,balanced,37.26140340169271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,1,power_law_1.01,6.72635498046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,1,power_law_1.01,2.7636287689208983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,1,balanced,4.567040125528972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,1,power_law_1.2,1.3528191566467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,1,balanced,4.615472157796224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,1,power_law_1.2,1.8591808319091796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,1,power_law_1.01,6.941683197021485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,1,balanced,4.670639991760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,1,power_law_1.01,0.8600064277648926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,1,power_law_1.2,3.475040054321289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,1,power_law_1.01,1.218496036529541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,1,balanced,4.767701466878255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,1,power_law_1.01,7.901158142089844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,1,power_law_1.2,0.14809600114822388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,1,balanced,4.932943979899089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,1,power_law_1.2,0.2569727897644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,1,power_law_1.2,0.47626237869262694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,1,balanced,5.096506754557292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,1,power_law_1.01,0.8501376152038574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,1,power_law_1.01,3.351603317260742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,1,power_law_1.01,0.9433088302612305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,1,power_law_1.01,9.190476989746093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,1,power_law_1.01,1.1374655723571778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,1,power_law_1.01,3.7637760162353517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,1,power_law_1.01,1.3495167732238769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,1,balanced,8.7304318745931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,1,power_law_1.2,3.8961280822753905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,1,power_law_1.01,1.7596223831176758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,1,power_law_1.01,4.2483264923095705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,1,power_law_1.01,2.0298303604125976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,1,power_law_1.01,2.2156864166259767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,1,power_law_1.2,0.7384064197540283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,1,power_law_1.2,4.256288146972656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,1,power_law_1.2,1.0766079902648926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,1,power_law_1.01,2.146124839782715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,1,balanced,8.97866121927897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,1,power_law_1.01,3.039014434814453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,1,power_law_1.01,10.629138946533203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,1,power_law_1.2,0.8769791603088379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,1,power_law_1.01,4.7178497314453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,1,power_law_1.2,0.9823424339294433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,1,power_law_1.01,2.263871955871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,1,power_law_1.2,1.1746368408203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,1,power_law_1.01,4.024185562133789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,1,power_law_1.2,1.3764096260070802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,1,power_law_1.01,5.866751861572266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,1,power_law_1.2,1.9184383392333983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,1,balanced,13.076666514078775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,1,power_law_1.01,12.63220443725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,1,power_law_1.01,6.858560180664062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,1,power_law_1.2,1.793062400817871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,1,power_law_1.2,4.543257522583008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,1,power_law_1.2,2.270521545410156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,1,balanced,13.59829839070638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,1,power_law_1.2,4.645702362060547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,1,power_law_1.2,2.3641664505004885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,1,power_law_1.01,9.48635482788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,1,power_law_1.2,2.4164735794067385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,1,power_law_1.01,16.410791015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,1,power_law_1.01,3.732563018798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,1,power_law_1.01,11.304774475097656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,1,balanced,21.5777587890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,1,power_law_1.01,3.7804927825927734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,1,power_law_1.01,3.8915775299072264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,1,power_law_1.2,4.624063873291016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,1,power_law_1.2,2.3629568099975584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,1,power_law_1.2,0.1620352029800415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,1,power_law_1.2,0.2828927993774414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,1,power_law_1.2,2.432307243347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,1,power_law_1.01,4.031552124023437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,1,power_law_1.01,19.843161010742186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,1,power_law_1.2,0.4998784065246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,1,power_law_1.01,15.882675170898438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,1,power_law_1.01,4.096992111206054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,1,power_law_1.2,3.0517759323120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,1,balanced,26.0546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,1,power_law_1.01,4.178169631958008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,1,power_law_1.2,3.9106494903564455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,1,power_law_1.01,4.372153472900391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,1,power_law_1.2,4.724576187133789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,1,power_law_1.01,4.561862564086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,1,power_law_1.2,0.7802048206329346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,1,power_law_1.2,4.986431884765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,1,power_law_1.2,2.359872055053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,1,power_law_1.2,1.0952128410339355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,1,power_law_1.01,20.45166015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,1,power_law_1.01,4.880460739135742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,2,balanced,0.10485333204269409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,2,balanced,0.1167680025100708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,2,balanced,0.17289066314697266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,2,balanced,0.1667733391125997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,2,balanced,0.16743467251459757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,2,balanced,0.16880534092585245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,2,balanced,0.2217173377672831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,2,balanced,0.22155733903249106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,1,power_law_1.2,5.220870590209961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,2,balanced,0.22873065869013467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,2,balanced,0.23522132635116577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,2,balanced,0.19711466630299887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,2,balanced,0.2032960057258606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,2,balanced,0.2058186729749044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,2,balanced,0.21772799889246622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,2,balanced,0.255514661471049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,2,balanced,0.2625653346379598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,1,power_law_1.01,5.575212860107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,2,balanced,0.2708853284517924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,2,balanced,0.33080534140268963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,2,balanced,0.3566506703694661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,2,balanced,0.5198560158411661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,2,balanced,0.6362826824188232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,1,power_law_1.2,5.292812728881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,2,balanced,0.9265759785970052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,1,power_law_1.01,27.250247192382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,2,balanced,1.1946773529052734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,2,balanced,1.7722400029500325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,1,power_law_1.01,6.310611343383789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,2,balanced,0.11020267009735107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,2,balanced,0.14849066734313965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,2,balanced,0.24848532676696777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,2,balanced,2.262154738108317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,1,balanced,38.470133463541664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,1,power_law_1.2,2.406175994873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,1,power_law_1.2,5.292044830322266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,1,power_law_1.2,1.9753023147583009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,2,balanced,3.4835999806722007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,1,power_law_1.2,2.5277055740356444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,1,power_law_1.2,2.158393669128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,1,power_law_1.01,7.717926025390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,2,balanced,4.754570643107097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,1,power_law_1.2,2.598355293273926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,1,power_law_1.2,5.539519882202148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,1,power_law_1.2,2.231328010559082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,1,power_law_1.2,2.742380714416504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,1,power_law_1.2,2.7583168029785154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,1,power_law_1.01,9.128217315673828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,1,power_law_1.2,2.8645631790161135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,2,balanced,0.24787733952204385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,2,balanced,0.24893333514531454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,2,balanced,0.24944533904393515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,1,power_law_1.01,11.98427505493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,1,power_law_1.01,35.52122192382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,1,power_law_1.2,6.616716766357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,1,balanced,47.449788411458336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,1,power_law_1.2,7.133702087402344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,2,balanced,0.12738666931788126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,2,balanced,0.1444533367951711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,2,balanced,0.19299199183781943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,1,power_law_1.2,3.695916748046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,1,power_law_1.01,14.863551330566406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,1,power_law_1.2,3.787519836425781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,1,power_law_1.2,8.229682922363281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,2,balanced,0.2856480081876119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,1,power_law_1.2,4.019961547851563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,2,balanced,0.2864266633987427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,1,power_law_1.2,3.4330753326416015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,1,power_law_1.2,4.059257507324219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,1,power_law_1.2,9.699104309082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,1,power_law_1.2,3.8512958526611327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,1,power_law_1.2,4.14334716796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,1,power_law_1.2,4.385715103149414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,2,power_law_1.01,0.11603200435638428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,2,power_law_1.01,0.1494271993637085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,1,power_law_1.01,20.5172607421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,2,power_law_1.01,0.20266239643096923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,1,power_law_1.2,4.406054306030273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,1,power_law_1.2,4.823680114746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,1,power_law_1.2,11.13544921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,2,balanced,0.19364267587661743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,2,balanced,0.19913599888483682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,1,power_law_1.2,4.591308975219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,1,power_law_1.2,6.004441452026367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,1,power_law_1.2,4.75860481262207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,2,balanced,0.3352320194244385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,2,balanced,0.3363093137741089
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,2,balanced,0.04278933505217234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,2,balanced,0.05964266757170359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,2,balanced,0.0846560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,2,balanced,0.08661333719889323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,2,balanced,0.08921066919962566
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,2,balanced,0.08989333113034566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,1,power_law_1.2,12.8944580078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,1,power_law_1.2,7.349369812011719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,1,power_law_1.2,5.0967552185058596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,2,balanced,0.09126399954160054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,2,power_law_1.01,0.23877758979797364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,1,power_law_1.2,5.778662490844726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,2,power_law_1.01,0.250163197517395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,2,power_law_1.01,0.27100799083709715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,1,power_law_1.01,26.044805908203124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,1,power_law_1.2,9.405510711669923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,2,balanced,0.09305066863695781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,2,balanced,0.09480533003807068
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,2,balanced,0.09442133704821269
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,2,balanced,0.09611200292905171
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,2,balanced,0.12317867080370586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,2,balanced,0.10956266522407532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,2,balanced,0.10544000069300334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,2,balanced,0.26239466667175293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,2,balanced,0.17951999107996622
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,2,balanced,0.11411733428637187
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,2,balanced,0.25960532824198407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,1,power_law_1.2,16.50948486328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,2,balanced,0.163290669520696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,2,balanced,0.22040534019470215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,2,balanced,0.275711993376414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,2,balanced,0.38411200046539307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,2,balanced,0.48606399695078534
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,2,balanced,0.7096319993336996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,2,balanced,0.21584532658259073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,2,balanced,0.21861332654953003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,1,power_law_1.2,6.481843566894531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,2,balanced,0.20882133642832437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,1,power_law_1.2,11.51778564453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,2,balanced,0.9297653039296468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,2,balanced,1.3692213694254558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,1,power_law_1.2,7.900479888916015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,2,balanced,1.811845302581787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,2,power_law_1.01,0.09630079865455628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,2,power_law_1.01,0.10245120525360107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,2,power_law_1.01,0.12163840532302857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,2,power_law_1.01,0.15813759565353394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,2,power_law_1.01,0.16179840564727782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,2,power_law_1.01,0.2948352098464966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,2,power_law_1.01,0.16993279457092286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,2,power_law_1.01,0.29910399913787844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,2,power_law_1.01,0.2145792007446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,2,power_law_1.01,0.2421247959136963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,2,power_law_1.01,0.2388927936553955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,2,power_law_1.01,0.24903039932250975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,1,power_law_1.2,20.560665893554688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,2,power_law_1.01,0.20660479068756105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,2,power_law_1.01,0.2182080030441284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,1,power_law_1.2,9.35338897705078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,2,power_law_1.01,0.22990720272064208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,2,balanced,0.33195199569066364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,2,power_law_1.01,0.24979839324951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,2,power_law_1.01,0.295795202255249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,2,power_law_1.01,0.3158463954925537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,1,power_law_1.01,36.93116760253906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,2,power_law_1.01,0.3801599979400635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,2,power_law_1.01,0.36735360622406005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,2,power_law_1.01,0.4738111972808838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,1,power_law_1.2,12.142784118652344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,2,power_law_1.01,0.6332287788391113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,1,power_law_1.2,15.58721923828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,2,power_law_1.01,0.35250558853149416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,2,power_law_1.01,0.39472639560699463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,1,power_law_1.2,14.941127014160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,1,power_law_1.2,27.290682983398437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,2,power_law_1.01,0.8498047828674317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,2,balanced,0.3376213312149048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,2,balanced,0.33825600147247314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,2,balanced,0.3535999854405721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,2,balanced,0.408517320950826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,2,balanced,0.42022931575775146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,2,balanced,0.41837334632873535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,2,power_law_1.01,1.1303359985351562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,1,power_law_1.2,20.409426879882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,1,power_law_1.01,47.51767578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,2,power_law_1.01,0.1332159996032715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,2,power_law_1.01,0.15011199712753295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,2,power_law_1.01,0.1771008014678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,2,balanced,0.32472532987594604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,2,balanced,0.3261386752128601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,2,balanced,0.33081066608428955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,2,balanced,0.34162668387095135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,2,power_law_1.01,0.34332799911499023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,2,balanced,0.34594134489695233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,2,balanced,0.3661493460337321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,2,balanced,0.4761279821395874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,2,balanced,0.49432531992594403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,2,balanced,0.5044586658477783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,2,balanced,0.7647253672281901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,2,balanced,0.9081172943115234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,2,power_law_1.01,0.1907647967338562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,2,balanced,1.2957066694895427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,2,power_law_1.01,0.19793280363082885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,2,power_law_1.01,0.21385600566864013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,2,balanced,1.6991465886433919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,1,power_law_1.2,35.89088745117188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,2,balanced,2.511626720428467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,2,balanced,0.522437334060669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,2,balanced,0.5566773414611816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,2,power_law_1.01,1.5298368453979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,2,balanced,3.1853920618693032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,2,balanced,0.8129706382751465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,2,balanced,0.9846399625142416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,2,power_law_1.01,0.22767999172210693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,2,power_law_1.01,0.24034559726715088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,2,balanced,1.450602690378825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,2,power_law_1.01,0.3231040000915527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,1,power_law_1.2,25.785153198242188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,2,balanced,4.807109196980794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,2,power_law_1.01,0.331110405921936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,2,balanced,1.879327932993571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,2,power_law_1.01,0.3448319911956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,2,power_law_1.01,0.37406721115112307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,2,balanced,2.8534132639567056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,2,power_law_1.01,0.40561280250549314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,2,power_law_1.01,0.4563007831573486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,2,balanced,6.15066655476888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,2,power_law_1.01,1.9945344924926758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,2,power_law_1.01,0.5503488063812256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,2,balanced,3.66321595509847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,2,power_law_1.01,0.6594304084777832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,2,power_law_1.01,0.7609344005584717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,2,balanced,6.007157643636067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,2,balanced,9.294048309326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,2,power_law_1.01,0.3684799909591675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,2,power_law_1.01,0.9757311820983887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,2,power_law_1.01,0.39495038986206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,2,power_law_1.01,0.43674240112304685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,1,power_law_1.2,19.94622039794922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,2,balanced,8.108144124348959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,2,power_law_1.01,0.4830912113189697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,2,power_law_1.2,0.10597120523452759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,2,power_law_1.2,0.14016000032424927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,2,power_law_1.01,1.3425151824951171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,2,power_law_1.2,0.14528000354766846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,2,power_law_1.2,0.16901760101318358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,2,power_law_1.01,0.5131840229034423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,2,power_law_1.2,0.16649600267410278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,2,power_law_1.2,0.16714240312576295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,2,balanced,12.260320027669271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,2,power_law_1.2,0.21733760833740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,2,power_law_1.2,0.22314879894256592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,2,power_law_1.2,0.13489279747009278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,2,power_law_1.01,0.6736000061035157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,2,power_law_1.2,0.15777280330657958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,2,power_law_1.2,0.11973119974136352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,2,power_law_1.2,0.1819391965866089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,2,power_law_1.2,0.14959360361099244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,2,power_law_1.2,0.19007999897003175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,2,power_law_1.2,0.18133120536804198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,2,power_law_1.2,0.19814399480819703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,2,power_law_1.2,0.24623360633850097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,2,power_law_1.2,0.21215999126434326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,2,power_law_1.2,0.24931199550628663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,2,power_law_1.2,0.22104320526123047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,2,power_law_1.2,0.28460800647735596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,2,power_law_1.2,0.24697599411010743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,2,power_law_1.2,0.30227200984954833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,2,power_law_1.2,0.3258687973022461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,2,power_law_1.2,0.3029247999191284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,2,power_law_1.2,0.23625600337982178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,2,power_law_1.2,0.3310976028442383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,2,power_law_1.2,0.36453120708465575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,2,power_law_1.2,0.2650559902191162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,2,power_law_1.2,0.35145599842071534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,2,power_law_1.2,0.3727423906326294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,2,power_law_1.2,0.20189440250396729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,2,power_law_1.2,0.3799040079116821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,2,power_law_1.2,0.34932479858398435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,2,power_law_1.2,0.2174015998840332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,2,power_law_1.2,0.38664960861206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,2,power_law_1.2,0.23615360260009766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,2,power_law_1.2,0.2439743995666504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,2,power_law_1.01,1.6478143692016602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,2,power_law_1.2,0.4463359832763672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,1,power_law_1.2,36.47427062988281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,2,power_law_1.2,0.28048000335693357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,2,power_law_1.2,0.5508607864379883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,2,power_law_1.2,0.6417727947235108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,2,power_law_1.2,0.31775360107421874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,2,power_law_1.2,0.39045760631561277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,2,power_law_1.2,0.747321605682373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,2,power_law_1.01,2.587129592895508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,2,power_law_1.2,0.38355839252471924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,2,power_law_1.2,0.3976448059082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,2,power_law_1.01,0.05129600167274475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,2,power_law_1.2,0.43920001983642576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,2,power_law_1.2,0.94399995803833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,2,power_law_1.01,0.06273919939994813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,2,power_law_1.2,0.4790976047515869
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,2,power_law_1.01,0.0651199996471405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,2,power_law_1.01,2.2271488189697264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,2,power_law_1.01,0.08401920199394226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,2,power_law_1.01,0.08792960047721862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,2,power_law_1.2,0.5489344120025634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,2,power_law_1.2,1.270470428466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,2,power_law_1.01,0.10170880556106568
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,2,power_law_1.01,0.099481600522995
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,2,power_law_1.01,0.10423680543899536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,2,power_law_1.01,0.10621440410614014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,2,power_law_1.01,0.107315194606781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,2,power_law_1.01,0.1102720022201538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,2,power_law_1.2,0.6402624130249024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,2,power_law_1.01,0.13027199506759643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,2,power_law_1.2,1.6429439544677735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,2,power_law_1.01,0.12328959703445434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,2,power_law_1.01,0.13925119638442993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,2,power_law_1.2,0.3908735990524292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,2,power_law_1.01,0.1549183964729309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,2,power_law_1.2,0.6840320110321045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,2,power_law_1.01,0.14303359985351563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,2,power_law_1.01,3.270969772338867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,2,power_law_1.01,0.1800063967704773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,2,power_law_1.2,0.4842688083648682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,2,power_law_1.01,0.6634624004364014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,2,power_law_1.01,0.7661888122558593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,2,power_law_1.2,0.8146368026733398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,2,power_law_1.01,0.2149951934814453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,2,power_law_1.2,0.6310272216796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,2,power_law_1.2,2.244838333129883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,2,power_law_1.01,0.27299840450286866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,2,power_law_1.2,1.1133503913879395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,2,power_law_1.01,0.9704256057739258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,2,power_law_1.2,0.8104512214660644
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,2,power_law_1.01,0.3254528045654297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,1,power_law_1.2,46.858203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,2,power_law_1.01,1.2648192405700684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,2,power_law_1.01,0.36954240798950194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,2,power_law_1.2,1.1284095764160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,2,power_law_1.2,3.7107391357421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,2,power_law_1.01,1.775712013244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,2,power_law_1.2,1.4029248237609864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,2,power_law_1.2,3.8227073669433596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,2,power_law_1.01,0.44929919242858884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,2,power_law_1.01,3.760755157470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,2,power_law_1.01,2.421235275268555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,2,power_law_1.2,1.4987775802612304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,2,power_law_1.01,3.8354686737060546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,2,power_law_1.2,1.807027244567871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,2,power_law_1.01,4.337440109252929
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,2,power_law_1.2,0.04737919867038727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,2,power_law_1.2,0.06234880089759827
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,2,power_law_1.2,0.06228479743003845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,2,power_law_1.2,0.08485760092735291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,2,power_law_1.2,0.0963648021221161
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,2,power_law_1.2,0.1027519941329956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,2,power_law_1.2,0.10063999891281128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,2,power_law_1.2,0.10524159669876099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,2,power_law_1.01,6.011750411987305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,2,power_law_1.2,0.10907520055770874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,2,power_law_1.2,0.1109503984451294
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,2,power_law_1.2,0.11822079420089722
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,2,power_law_1.2,0.1289728045463562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,2,power_law_1.2,0.1277119994163513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,2,power_law_1.2,0.14151040315628052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,2,power_law_1.2,0.1454208016395569
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,2,power_law_1.2,0.18833919763565063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,2,power_law_1.01,0.7517632007598877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,2,power_law_1.2,0.1547775983810425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,2,power_law_1.2,0.21412479877471924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,2,power_law_1.2,2.547916793823242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,2,power_law_1.2,0.21527040004730225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,2,power_law_1.2,0.2971839904785156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,2,power_law_1.2,5.7522624969482425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,2,power_law_1.2,2.025164794921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,2,power_law_1.2,0.354367995262146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,2,power_law_1.01,7.621971130371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,2,power_law_1.01,4.408160018920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,2,power_law_1.01,1.0462719917297363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,2,power_law_1.01,6.257183837890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,2,power_law_1.2,0.539353609085083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,2,1,balanced,0.09052800138791402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,2,power_law_1.2,3.801542282104492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,2,power_law_1.2,8.470079803466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,2,1,balanced,0.1357866624991099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,2,1,balanced,0.25274133682250977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,2,power_law_1.2,0.6765247821807862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,2,power_law_1.01,8.59423370361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,2,power_law_1.01,1.2758655548095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,2,1,balanced,0.2500266631444295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,2,1,balanced,0.25616000096003216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,2,1,balanced,0.25576533873875934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,2,1,balanced,0.2539253234863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,2,1,balanced,0.25600532690684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,2,power_law_1.2,2.7484479904174806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,2,power_law_1.01,1.947065544128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,2,power_law_1.2,0.8897024154663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,2,power_law_1.01,12.406758117675782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,2,power_law_1.2,4.297196960449218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,2,1,balanced,0.27591999371846515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,2,1,balanced,0.2834239999453227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,2,power_law_1.2,11.479334259033203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,2,power_law_1.01,17.029434204101562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,2,1,balanced,0.06364266574382782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,2,1,balanced,0.09089066584904988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,2,1,balanced,0.14011200269063315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,2,1,balanced,0.1397599975268046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,2,1,balanced,0.14113600055376688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,2,1,balanced,0.14808533589045206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,2,1,balanced,0.17412267128626505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,2,1,balanced,0.1862293283144633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,2,1,balanced,0.1722666621208191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,2,1,balanced,0.1881706714630127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,2,1,balanced,0.16475199659665427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,2,1,balanced,0.16878400246302286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,2,1,balanced,0.17174400885899863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,2,1,balanced,0.17664533853530884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,2,1,balanced,0.2323253353436788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,2,1,balanced,0.23795199394226074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,2,1,balanced,0.237119992574056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,2,power_law_1.01,2.257702445983887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,2,1,balanced,0.30054400364557904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,2,1,balanced,0.3173919916152954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,2,1,balanced,0.4219626585642497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,2,1,balanced,0.5275413195292155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,2,1,balanced,0.762277364730835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,2,1,balanced,0.9543039798736572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,2,1,balanced,1.4211680094401042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,2,1,balanced,1.818021297454834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,2,1,balanced,2.7644532521565757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,2,1,balanced,3.892319997151693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,2,1,balanced,0.2812160054842631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,2,power_law_1.2,4.152268981933593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,2,1,balanced,0.08541867136955261
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,2,1,balanced,0.041135999063650765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,2,1,balanced,0.05789333085219065
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,2,1,balanced,0.08521599570910136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,2,1,balanced,0.08556800087292989
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,2,1,balanced,0.08699199557304382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,2,1,balanced,0.0899733304977417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,2,1,balanced,0.0988213320573171
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,2,1,balanced,0.10096533099810283
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,2,1,balanced,0.10128532846768697
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,2,1,balanced,0.10156266887982686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,2,1,balanced,0.10499733686447144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,2,1,balanced,0.108815997838974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,2,1,balanced,0.2848106622695923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,2,1,balanced,0.10975999633471172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,2,1,balanced,0.2874346574147542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,2,1,balanced,0.11416533589363098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,2,1,balanced,0.29391467571258545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,2,1,balanced,0.39658133188883465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,2,1,balanced,0.39660267035166424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,2,1,balanced,0.39631466070810956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,2,1,balanced,0.10855467120806377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,2,1,balanced,0.15982932845751444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,2,1,balanced,0.1991680065790812
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,2,1,balanced,0.16516266266504923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,2,1,balanced,0.1302826702594757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,2,1,balanced,0.32284800211588544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,2,1,balanced,0.16198933124542236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,2,1,balanced,0.21397332350413004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,2,1,balanced,0.2614026665687561
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,2,1,balanced,0.36708799997965497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,2,1,balanced,0.47275733947753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,2,1,balanced,0.6861493587493896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,2,1,balanced,0.9028159777323405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,2,1,balanced,1.3374773661295574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,2,1,balanced,1.7690025965372722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,2,1,power_law_1.01,0.08543999791145325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,2,power_law_1.01,10.662739562988282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,2,1,balanced,0.5045119921366373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,2,1,balanced,0.15825066963831583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,2,1,balanced,0.15917332967122397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,2,1,balanced,0.524069348971049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,2,1,power_law_1.01,0.11373440027236939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,2,1,power_law_1.01,0.13889919519424437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,2,power_law_1.2,1.2943807601928712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,2,1,power_law_1.01,0.05920000076293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,2,1,balanced,0.1676479975382487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,2,1,balanced,0.16945600509643555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,2,1,balanced,0.169813334941864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,2,1,balanced,0.7089493274688721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,2,1,balanced,0.882693370183309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,2,1,balanced,1.278608004252116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,2,1,balanced,1.5963466962178547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,2,1,balanced,2.478991985321045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,2,1,balanced,3.1193014780680337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,2,1,power_law_1.01,0.07997440099716187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,2,1,power_law_1.01,0.10024319887161255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,2,1,power_law_1.01,0.2325376033782959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,2,1,power_law_1.01,0.25326080322265626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,2,1,power_law_1.01,0.26350719928741456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,2,1,balanced,4.745375951131185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,2,1,balanced,6.612016042073567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,2,1,power_law_1.01,0.09178879857063293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,2,power_law_1.2,7.113107299804687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,2,1,power_law_1.01,0.13941119909286498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,2,1,power_law_1.01,0.13916800022125245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,2,1,power_law_1.01,0.14878079891204835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,2,1,power_law_1.01,0.2525376081466675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,2,1,power_law_1.01,0.0996288001537323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,2,1,power_law_1.01,0.113811194896698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,2,1,power_law_1.01,0.2561984062194824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,2,1,balanced,0.25518399477005005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,2,1,balanced,0.25787200530370075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,2,1,balanced,0.26119999090830487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,2,1,balanced,0.2664320071538289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,2,1,balanced,0.2734666665395101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,2,1,balanced,0.2807893355687459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,2,1,balanced,0.4401386578877767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,2,1,balanced,0.44993066787719727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,2,1,balanced,0.45371198654174805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,2,1,balanced,0.6326133410135905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,2,1,balanced,0.8085866769154867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,2,1,balanced,1.1628959973653157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,2,1,balanced,1.4667305946350098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,2,1,balanced,2.1691039403279624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,2,1,power_law_1.01,0.15754239559173583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,2,1,power_law_1.01,0.15971839427947998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,2,1,power_law_1.01,0.1716863989830017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,2,1,power_law_1.01,0.18520959615707397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,2,1,balanced,2.7939891815185547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,2,1,balanced,4.168149312337239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,2,1,balanced,5.3241119384765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,2,1,power_law_1.01,0.2916928052902222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,2,1,power_law_1.01,0.29786880016326905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,2,1,power_law_1.01,0.16227840185165404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,2,1,power_law_1.01,0.1745151996612549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,2,1,power_law_1.01,0.18887679576873778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,2,power_law_1.2,16.139276123046876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,2,1,balanced,8.031493504842123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,2,power_law_1.2,1.6331775665283204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,2,1,balanced,10.529120127360025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,2,1,power_law_1.2,0.08487679958343505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,2,1,power_law_1.2,0.11845120191574096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,2,1,power_law_1.2,0.14470399618148805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,2,1,power_law_1.2,0.2342911958694458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,2,1,power_law_1.2,0.25358080863952637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,2,1,power_law_1.2,0.26306560039520266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,2,1,power_law_1.2,0.26064000129699705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,2,1,power_law_1.2,0.2739903926849365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,2,1,power_law_1.2,0.28769280910491946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,2,1,power_law_1.2,0.30845439434051514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,2,1,power_law_1.2,0.27743360996246336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,2,1,power_law_1.2,0.2902976036071777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,2,1,power_law_1.01,0.19984639883041383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,2,power_law_1.2,5.4538623809814455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,2,1,power_law_1.2,0.2994688034057617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,2,1,power_law_1.2,0.06423680186271667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,2,1,power_law_1.2,0.07964159846305847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,2,1,power_law_1.2,0.3247423887252808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,2,1,power_law_1.2,0.11101440191268921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,2,1,power_law_1.2,0.13987200260162352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,2,1,power_law_1.2,0.14011520147323608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,2,1,power_law_1.2,0.15351680517196656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,2,1,power_law_1.2,0.1851456046104431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,2,1,power_law_1.2,0.3850752115249634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,2,1,power_law_1.2,0.19223040342330933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,2,1,power_law_1.2,0.2009216070175171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,2,1,power_law_1.2,0.4133120059967041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,2,1,power_law_1.01,0.210265588760376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,2,1,power_law_1.2,0.2099008083343506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,2,1,power_law_1.2,0.16222720146179198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,2,1,power_law_1.2,0.16899839639663697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,2,1,power_law_1.2,0.17585279941558837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,2,1,power_law_1.01,0.04054400026798248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,2,1,power_law_1.2,0.20470399856567384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,2,1,power_law_1.01,0.2535552024841309
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,2,1,power_law_1.01,0.05398399829864502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,2,1,power_law_1.01,0.24899840354919434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,2,1,power_law_1.01,0.05891839861869812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,2,1,power_law_1.01,0.2831615924835205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,2,1,power_law_1.01,0.31736319065093993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,2,1,power_law_1.01,0.3449408054351807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,2,power_law_1.2,11.64961929321289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,2,1,power_law_1.01,0.38559999465942385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,2,1,power_law_1.01,0.40766081809997556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,2,1,power_law_1.01,0.28185598850250243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,2,1,power_law_1.2,0.23008639812469484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,2,1,power_law_1.01,0.4512959957122803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,2,1,power_law_1.01,0.07841280102729797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,2,1,power_law_1.01,0.08244479894638061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,2,1,power_law_1.01,0.5483456134796143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,2,1,power_law_1.01,0.0898688018321991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,2,1,power_law_1.2,0.24208641052246094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,2,1,power_law_1.01,0.09921280145645142
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,2,1,power_law_1.01,0.11112320423126221
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,2,1,power_law_1.01,0.0995199978351593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,2,1,power_law_1.01,0.10617599487304688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,2,1,power_law_1.01,0.7109119892120361
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,2,1,power_law_1.01,0.11624959707260132
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,2,1,power_law_1.01,0.10986239910125732
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,2,1,power_law_1.01,0.115174400806427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,2,1,power_law_1.01,0.8681792259216309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,2,1,power_law_1.2,0.4472640037536621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,2,1,power_law_1.01,0.1295359969139099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,2,1,power_law_1.01,0.12923519611358641
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,2,1,power_law_1.01,0.13603199720382692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,2,1,power_law_1.2,0.26301438808441163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,2,1,power_law_1.01,0.15523840188980104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,2,1,power_law_1.2,0.08488320112228394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,2,1,power_law_1.01,1.2059840202331542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,2,1,power_law_1.01,0.19392640590667726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,2,1,power_law_1.2,0.2984256029129028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,2,1,power_law_1.2,0.5065023899078369
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,2,1,power_law_1.01,0.18896000385284423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,2,1,power_law_1.01,1.5441920280456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,2,1,power_law_1.01,0.23923840522766113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,2,1,power_law_1.2,0.3561088085174561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,2,1,power_law_1.2,0.6320127964019775
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,2,power_law_1.2,2.8842304229736326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,2,1,power_law_1.01,0.29171841144561766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,2,1,power_law_1.01,0.16437760591506959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,2,1,power_law_1.01,0.29404799938201903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,2,1,power_law_1.01,0.325382399559021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,2,1,power_law_1.01,0.38778879642486574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,2,1,power_law_1.2,0.45027837753295896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,2,1,power_law_1.01,0.4182079792022705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,2,1,power_law_1.01,0.29205119609832764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,2,1,power_law_1.01,0.45261440277099607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,2,1,power_law_1.01,2.2016063690185548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,2,1,power_law_1.01,0.4045695781707764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,2,1,power_law_1.2,0.7746560096740722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,2,1,power_law_1.2,0.10019199848175049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,2,1,power_law_1.2,0.5535615921020508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,2,1,power_law_1.2,0.10803200006484985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,2,1,power_law_1.2,0.04043520092964172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,2,1,power_law_1.2,0.05242879986763001
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,2,1,power_law_1.2,0.0616320013999939
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,2,1,power_law_1.2,0.08264319896697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,2,1,power_law_1.2,0.08334720134735107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,2,1,power_law_1.2,0.0921343982219696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,2,1,power_law_1.2,0.10087039470672607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,2,1,power_law_1.01,0.1678015947341919
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,2,1,power_law_1.2,0.1118783950805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,2,1,power_law_1.01,0.1719040036201477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,2,1,power_law_1.2,0.10090880393981934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,2,1,power_law_1.01,0.18754559755325317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,2,1,power_law_1.2,0.10565119981765747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,2,1,power_law_1.01,0.2231231927871704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,2,1,power_law_1.2,0.9601984024047852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,2,1,power_law_1.2,0.1141759991645813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,2,1,power_law_1.01,0.24348800182342528
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,2,1,power_law_1.2,0.1085312008857727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,2,1,power_law_1.2,0.11563520431518555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,2,1,power_law_1.01,0.2625279903411865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,2,1,power_law_1.2,0.12929919958114625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,2,1,power_law_1.2,0.7773952007293701
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,2,1,power_law_1.2,0.12737280130386353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,2,1,power_law_1.2,0.13626879453659058
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,2,1,power_law_1.01,0.5033919811248779
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,2,1,power_law_1.2,0.15442559719085694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,2,1,power_law_1.01,0.5156223773956299
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,2,1,power_law_1.2,0.16222720146179198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,2,1,power_law_1.01,2.8462080001831054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,2,1,power_law_1.01,0.6364543914794922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,2,1,power_law_1.2,1.3346624374389648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,2,1,power_law_1.2,0.1565951943397522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,2,1,power_law_1.2,0.1593984007835388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,2,1,power_law_1.2,0.21173760890960694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,2,1,power_law_1.01,4.1622974395751955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,2,1,power_law_1.2,0.2467519998550415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,2,1,power_law_1.01,0.7111167907714844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,2,1,power_law_1.2,0.29711360931396485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,2,1,power_law_1.2,1.000115203857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,2,1,power_law_1.01,0.30456318855285647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,2,1,power_law_1.01,0.35868799686431885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,2,1,power_law_1.2,1.7097791671752929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,2,1,power_law_1.2,0.16276479959487916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,2,1,power_law_1.2,0.1728000044822693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,2,1,power_law_1.2,0.18983039855957032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,2,1,power_law_1.01,0.7867584228515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,2,1,power_law_1.01,0.9587776184082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,2,1,power_law_1.2,0.42471041679382326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,2,1,power_law_1.01,1.3424832344055175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,2,1,power_law_1.2,0.5081535816192627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,2,1,power_law_1.01,0.9516351699829102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,2,1,power_law_1.2,2.455865669250488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,2,1,power_law_1.01,5.488639831542969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,2,1,power_law_1.01,0.45576958656311034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,2,1,power_law_1.01,0.5446591854095459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,2,1,power_law_1.2,3.2289215087890626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,2,1,power_law_1.01,0.7567103862762451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,2,1,power_law_1.2,0.25345280170440676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,2,1,power_law_1.2,0.26711039543151854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,2,1,power_law_1.01,1.7167743682861327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,2,1,power_law_1.2,1.4334783554077148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,2,1,power_law_1.01,0.9866751670837403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,2,1,power_law_1.2,0.28354558944702146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,2,1,power_law_1.2,0.334335994720459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,2,1,power_law_1.2,0.3442176103591919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,2,1,power_law_1.2,0.3683840036392212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,2,1,power_law_1.2,0.40327038764953616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,2,1,power_law_1.2,0.4517055988311768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,2,1,power_law_1.2,0.7372159957885742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,2,1,power_law_1.2,0.5476992130279541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,2,1,power_law_1.2,0.7258751869201661
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,2,1,power_law_1.01,1.3622591972351075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,2,1,power_law_1.2,0.8729087829589843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,2,1,power_law_1.01,1.4052032470703124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,2,1,power_law_1.2,1.8641855239868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,2,1,power_law_1.2,1.2001728057861327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,2,1,power_law_1.01,1.8702592849731445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,2,1,power_law_1.01,2.461222457885742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,2,1,power_law_1.01,8.071250915527344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,2,1,power_law_1.2,1.5323967933654785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,2,1,power_law_1.2,2.2014463424682615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,2,1,power_law_1.2,2.744044876098633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,2,1,power_law_1.01,2.7095359802246093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,2,1,power_law_1.01,3.231999969482422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,2,1,power_law_1.01,4.799615859985352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,2,1,power_law_1.01,1.86297607421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,2,balanced,0.15477866927782694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,2,balanced,0.2254026730855306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,2,balanced,0.3733706474304199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,2,balanced,0.3737226724624634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,2,balanced,0.3768320083618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,2,balanced,0.3781386613845825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,2,balanced,0.4347626765569051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,2,balanced,0.4334719975789388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,2,balanced,0.4805599848429362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,2,balanced,0.48127468427022296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,2,balanced,0.47333331902821857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,2,balanced,0.48155732949574787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,2,balanced,0.4854453404744466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,2,balanced,0.4974079926808675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,2,balanced,0.6778026421864828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,2,balanced,0.6898079713185629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,2,balanced,0.6988960107167562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,2,balanced,0.8589173158009847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,2,1,power_law_1.2,0.9371904373168946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,2,balanced,0.8985599676767985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,2,balanced,1.1833706696828206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,2,1,power_law_1.2,2.848678398132324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,2,balanced,1.5821654001871746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,2,balanced,2.2501279513041177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,2,balanced,3.016677220662435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,2,balanced,4.378373463948567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,2,1,power_law_1.2,5.015552139282226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,2,balanced,5.956437428792317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,2,balanced,10.034613291422525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,2,balanced,13.600026448567709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,2,balanced,0.142277330160141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,2,balanced,0.17546667655309042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,2,balanced,0.2898719906806946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,2,balanced,0.2903093298276265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,2,balanced,0.2925279935201009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,2,balanced,0.2925279935201009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,2,balanced,0.3312373359998067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,2,balanced,0.3348960081736247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,2,balanced,0.310970664024353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,2,balanced,0.30107200145721436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,2,balanced,0.2711840073267619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,2,balanced,0.2763520081837972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,2,balanced,0.28412266572316486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,2,1,power_law_1.2,4.163564682006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,2,balanced,0.2964479923248291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,2,balanced,0.39185067017873126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,2,balanced,0.4028053283691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,2,balanced,0.40760000546773273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,2,balanced,0.5145706733067831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,2,balanced,0.5536479949951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,2,balanced,0.7176693280537924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,2,balanced,0.9391733010609945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,2,balanced,1.3431679407755535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,2,balanced,1.7731787363688152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,2,balanced,2.631253401438395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,2,balanced,3.481178601582845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,2,balanced,5.334442774454753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,2,balanced,7.344832102457683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,2,balanced,0.16679465770721436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,2,balanced,0.19229867060979208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,2,balanced,0.27187200387318927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,2,balanced,0.2725600004196167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,2,balanced,0.2757813334465027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,2,1,power_law_1.01,6.989318084716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,2,balanced,0.29199467102686566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,2,balanced,0.30031466484069824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,2,balanced,0.3161546587944031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,2,balanced,0.4388586680094401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,2,balanced,0.44152534008026123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,2,balanced,0.44657599925994873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,2,balanced,0.45947198073069256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,2,balanced,0.46486401557922363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,2,balanced,0.48908265431722003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,2,balanced,0.7815413475036621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,2,balanced,0.7988746960957845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,2,balanced,0.8182186285654703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,2,balanced,1.0664107004801433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,2,balanced,1.3901333808898926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,2,balanced,1.9533599217732747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,2,balanced,2.649834632873535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,2,balanced,3.806943893432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,2,balanced,5.018298784891765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,2,1,power_law_1.01,3.585728073120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,2,power_law_1.01,0.1761023998260498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,2,power_law_1.01,0.22705280780792236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,2,power_law_1.01,0.3023296117782593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,2,power_law_1.01,0.3723328113555908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,2,power_law_1.01,0.38259201049804686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,2,power_law_1.01,0.465177583694458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,2,balanced,7.450570424397786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,2,power_law_1.01,0.457260799407959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,2,power_law_1.01,0.48767361640930174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,2,power_law_1.01,0.5843200206756591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,2,power_law_1.01,0.5833471775054931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,2,power_law_1.01,0.5198016166687012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,2,power_law_1.01,0.5532800197601319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,2,power_law_1.01,0.5688384056091309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,2,balanced,9.603808085123697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,2,power_law_1.01,0.6383168220520019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,2,power_law_1.01,0.784876823425293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,2,power_law_1.01,0.9454272270202637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,2,1,power_law_1.01,10.678368377685548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,2,power_law_1.01,1.0032575607299805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,2,power_law_1.01,0.11710079908370971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,2,power_law_1.01,0.16181119680404663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,2,power_law_1.01,0.21578240394592285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,2,power_law_1.01,0.2363584041595459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,2,power_law_1.01,0.2458240032196045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,2,balanced,14.57593027750651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,2,power_law_1.01,0.26559360027313234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,2,power_law_1.01,0.32471039295196535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,2,power_law_1.01,0.34228479862213135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,2,power_law_1.01,0.3663424015045166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,2,power_law_1.01,1.0287296295166015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,2,power_law_1.01,0.40273280143737794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,2,power_law_1.01,0.27934720516204836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,2,power_law_1.01,1.1158080101013184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,2,1,power_law_1.2,5.452006530761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,2,power_law_1.01,0.3054527997970581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,2,power_law_1.01,0.3284991979598999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,2,power_law_1.01,0.35544960498809813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,2,power_law_1.01,0.41254401206970215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,2,power_law_1.01,0.4522751808166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,2,power_law_1.01,1.896460723876953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,2,power_law_1.01,0.5854976177215576
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,2,balanced,0.056976000467936196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,2,balanced,0.08235733211040497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,2,balanced,0.12919466694196066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,2,balanced,0.12930666406949362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,2,balanced,0.13099199533462524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,2,balanced,0.13367467125256857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,2,balanced,19.15104039510091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,2,power_law_1.01,0.5504960060119629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,2,power_law_1.01,0.1726591944694519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,2,power_law_1.01,0.20755839347839355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,2,power_law_1.01,0.24246399402618407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,2,power_law_1.01,0.6724160194396973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,2,power_law_1.01,0.26652159690856936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,2,power_law_1.01,0.26863999366760255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,2,power_law_1.01,2.2072128295898437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,2,1,power_law_1.2,1.3569984436035156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,2,power_law_1.01,0.31132800579071046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,2,balanced,0.1336426635583242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,2,balanced,0.13428800304730734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,2,power_law_1.01,0.34997758865356443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,2,balanced,0.13986666997273764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,2,balanced,0.14008000493049622
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,2,balanced,0.14286399881045023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,2,balanced,0.17103999853134155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,2,balanced,0.1623146633307139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,2,balanced,0.1602826714515686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,2,balanced,0.34486933549245197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,2,power_law_1.01,0.8699711799621582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,2,balanced,0.23904534180959067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,2,balanced,0.16819733381271362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,2,balanced,0.3696959813435872
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,2,balanced,0.25692800680796307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,2,balanced,0.3477439880371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,2,1,power_law_1.2,6.773516845703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,2,balanced,0.4612266620000203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,2,balanced,0.6530880133310953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,2,balanced,0.8867093722025553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,2,power_law_1.01,0.355020809173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,2,balanced,1.3338239987691243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,2,power_law_1.01,0.44458880424499514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,2,power_law_1.01,0.4656703948974609
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,2,balanced,1.7759413719177246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,2,power_law_1.01,0.4727168083190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,2,balanced,2.6582132975260415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,2,power_law_1.01,0.6097472190856934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,2,power_law_1.01,0.6183551788330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,2,power_law_1.01,0.7183040142059326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,2,balanced,3.5217386881510415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,2,power_law_1.2,0.20239360332489015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,2,power_law_1.01,1.1262016296386719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,2,power_law_1.2,0.2429503917694092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,2,power_law_1.2,0.3290112018585205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,2,power_law_1.01,0.8581567764282226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,2,power_law_1.2,0.38604800701141356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,2,power_law_1.2,0.36131839752197265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,2,power_law_1.2,0.4946303844451904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,2,power_law_1.01,1.7685056686401368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,2,power_law_1.01,0.9066431999206543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,2,power_law_1.2,0.45815038681030273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,2,power_law_1.01,3.155398368835449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,2,power_law_1.2,0.46822400093078614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,2,power_law_1.2,0.578656005859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,2,power_law_1.2,0.5800320148468018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,2,power_law_1.01,1.1292287826538085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,2,power_law_1.2,0.4996607780456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,2,power_law_1.2,0.5554304122924805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,2,power_law_1.2,0.623692798614502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,2,power_law_1.01,1.6546623229980468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,2,power_law_1.2,0.6680511951446533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,2,power_law_1.01,1.861427116394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,2,power_law_1.2,0.7904319763183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,2,power_law_1.01,2.214899253845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,2,power_law_1.2,0.8389311790466308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,2,power_law_1.2,1.1386176109313966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,2,power_law_1.01,2.999270439147949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,2,power_law_1.2,0.9644224166870117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,2,power_law_1.01,4.014444732666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,2,power_law_1.01,3.1238271713256838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,2,power_law_1.01,3.5188865661621094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,2,power_law_1.2,1.1245311737060546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,2,1,power_law_1.2,3.668947219848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,2,power_law_1.01,0.062412798404693604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,2,power_law_1.2,0.1404096007347107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,2,power_law_1.2,0.14661760330200196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,2,power_law_1.2,0.2123647928237915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,2,power_law_1.2,0.2505919933319092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,2,power_law_1.2,0.24976639747619628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,2,power_law_1.2,0.2832895994186401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,2,power_law_1.01,4.994028854370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,2,power_law_1.2,0.33255040645599365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,2,power_law_1.2,0.3491519927978516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,2,power_law_1.2,1.5447360038757325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,2,power_law_1.2,0.39648640155792236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,2,power_law_1.01,0.07900800108909607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,2,power_law_1.01,0.10124160051345825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,2,power_law_1.2,0.4335807800292969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,2,power_law_1.01,0.1213312029838562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,2,power_law_1.2,0.2821824073791504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,2,power_law_1.01,0.12316160202026367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,2,power_law_1.2,0.318284797668457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,2,power_law_1.01,0.166048002243042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,2,power_law_1.01,0.1450368046760559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,2,power_law_1.2,0.3406527996063232
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,2,power_law_1.01,0.18184959888458252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,2,power_law_1.01,0.15615999698638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,2,power_law_1.2,0.3610944032669067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,2,power_law_1.01,0.169760000705719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,2,power_law_1.01,0.19644800424575806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,2,power_law_1.01,0.20286080837249756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,2,power_law_1.01,0.19506560564041137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,2,power_law_1.01,6.556134033203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,2,power_law_1.01,0.22288000583648682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,2,power_law_1.01,0.29755520820617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,2,power_law_1.01,5.204064178466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,2,power_law_1.01,0.24673280715942383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,2,power_law_1.2,0.42170238494873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,2,power_law_1.01,0.2784575939178467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,2,power_law_1.2,0.48151679039001466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,2,power_law_1.01,0.31848959922790526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,2,power_law_1.2,2.270457649230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,2,power_law_1.2,0.5470335960388184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,2,power_law_1.01,0.33515520095825196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,2,power_law_1.2,0.5949376106262207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,2,power_law_1.01,0.5055168151855469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,2,power_law_1.2,0.6965888023376465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,2,power_law_1.2,2.8955583572387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,2,power_law_1.01,7.64901123046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,2,power_law_1.01,8.715692901611328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,2,power_law_1.01,0.6814911842346192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,2,power_law_1.2,1.019814395904541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,2,power_law_1.01,4.385209655761718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,2,power_law_1.2,3.7942783355712892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,2,power_law_1.2,1.2882623672485352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,2,power_law_1.01,0.9241215705871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,2,1,power_law_1.2,8.046822357177735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,2,power_law_1.01,1.2732607841491699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,2,power_law_1.2,1.7215295791625977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,2,power_law_1.2,2.2147199630737306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,2,power_law_1.01,11.251789093017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,2,power_law_1.2,6.1796928405761715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,2,power_law_1.01,1.8561023712158202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,2,power_law_1.01,14.21393280029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,2,1,power_law_1.2,1.882579231262207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,2,power_law_1.01,6.5055488586425785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,2,power_law_1.2,0.15946240425109864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,2,power_law_1.2,0.22112638950347902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,2,power_law_1.2,0.2037951946258545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,2,power_law_1.2,0.267737603187561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,2,power_law_1.2,0.26933119297027586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,2,power_law_1.2,0.3291071891784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,2,power_law_1.2,0.3370176076889038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,2,power_law_1.2,0.35761280059814454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,2,power_law_1.2,0.45163521766662595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,2,power_law_1.2,0.4744895935058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,2,power_law_1.2,3.3400768280029296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,2,power_law_1.2,0.48069119453430176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,2,power_law_1.2,0.5317887783050537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,2,power_law_1.2,0.5821504116058349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,2,power_law_1.2,0.714470386505127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,2,power_law_1.2,0.867347240447998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,2,power_law_1.2,0.9035967826843262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,2,power_law_1.2,1.0888192176818847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,2,power_law_1.01,2.3887935638427735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,2,power_law_1.2,1.6298431396484374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,2,power_law_1.2,8.128851318359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,2,power_law_1.2,1.930361557006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,2,power_law_1.2,2.5252031326293944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,2,power_law_1.01,3.5510398864746096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,2,power_law_1.2,3.47815055847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,2,power_law_1.01,16.011532592773438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,2,power_law_1.2,5.1212928771972654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,2,power_law_1.2,3.7870719909667967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,2,power_law_1.01,5.208204650878907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,2,1,balanced,0.15388799707094827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,2,1,balanced,0.24175999561945596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,2,1,balanced,0.40753066539764404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,2,1,balanced,0.4078986644744873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,2,1,balanced,0.4081173340479533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,2,1,balanced,0.410591999689738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,2,1,balanced,0.4747626781463623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,2,1,balanced,0.47147198518117267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,2,1,balanced,0.49259201685587567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,2,1,balanced,0.49321599801381427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,2,power_law_1.2,6.482093048095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,2,1,balanced,0.47574400901794434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,2,1,balanced,0.43724799156188965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,2,1,balanced,0.43908798694610596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,2,1,balanced,0.4482933282852173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,2,1,balanced,0.6059466600418091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,2,1,balanced,0.6144479910532633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,2,1,balanced,0.6230773528416952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,2,1,balanced,0.8070080280303955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,2,1,power_law_1.2,10.664300537109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,2,1,balanced,0.831216017405192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,2,1,balanced,1.122938632965088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,2,1,balanced,1.454213301340739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,2,1,balanced,2.0531360308329263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,2,1,balanced,2.758805274963379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,2,power_law_1.2,0.06298239827156067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,2,power_law_1.2,0.07893120050430298
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,2,power_law_1.2,0.07976319789886474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,2,1,balanced,4.110533396402995
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,2,power_law_1.2,0.12227200269699097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,2,power_law_1.2,0.138099205493927
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,2,power_law_1.2,0.17484159469604493
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,2,power_law_1.2,0.16771199703216552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,2,1,balanced,0.08396800359090169
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,2,power_law_1.2,0.18402559757232667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,2,power_law_1.2,0.15546239614486695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,2,1,balanced,5.602031707763672
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,2,power_law_1.2,0.17859840393066406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,2,power_law_1.2,0.19378559589385985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,2,power_law_1.2,0.18969600200653075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,2,power_law_1.01,9.849734497070312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,2,power_law_1.2,0.20300800800323487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,2,power_law_1.2,0.2275264024734497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,2,1,balanced,0.11276800433794658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,2,1,balanced,0.15431466698646545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,2,1,balanced,0.2448479930559794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,2,1,balanced,8.442373275756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,2,1,balanced,0.24338666598002115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,2,1,balanced,0.2410773237546285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,2,1,balanced,0.24744532505671182
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,2,power_law_1.2,0.2588799953460693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,2,1,balanced,0.2462773323059082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,2,1,balanced,0.24839999278386435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,2,1,balanced,0.4243946472803752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,2,power_law_1.2,0.23265280723571777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,2,1,balanced,0.42660800615946454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,2,1,balanced,0.42770131429036456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,2,1,balanced,0.4341973463694255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,2,1,balanced,0.43933331966400146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,2,power_law_1.2,0.260153603553772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,2,1,balanced,0.4508746862411499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,2,1,balanced,0.6664906740188599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,2,1,balanced,0.6795466740926107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,2,1,balanced,0.683509349822998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,2,1,balanced,0.9737706979115804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,2,power_law_1.2,0.31921279430389404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,2,1,balanced,1.2675039768218994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,2,1,balanced,12.175248463948568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,2,power_law_1.01,21.612045288085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,2,1,balanced,1.8076000213623047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,2,power_law_1.2,0.4069056034088135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,2,1,balanced,2.368607997894287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,2,1,balanced,3.48307736714681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,2,1,balanced,0.1356106698513031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,2,1,balanced,0.23529066642125449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,2,1,balanced,4.530330657958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,2,power_law_1.2,0.5868864059448242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,2,1,balanced,6.7328535715738935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,2,1,balanced,8.604607899983725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,2,power_law_1.2,0.6955391883850097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,2,power_law_1.2,10.083071899414062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,2,1,balanced,0.23251734177271524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,2,1,balanced,0.23099199930826822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,2,1,balanced,0.2376799980799357
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,2,1,balanced,0.053818667928377785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,2,1,balanced,0.07898133496443431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,2,1,balanced,0.12544000148773193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,2,1,balanced,0.12568533420562744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,2,1,balanced,0.12729600071907043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,2,1,balanced,0.12852266430854797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,2,1,balanced,0.14492799838383993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,2,1,balanced,0.1481706698735555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,2,1,balanced,0.14805866281191507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,2,1,balanced,0.1507200002670288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,2,1,balanced,0.15235732992490134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,2,1,balanced,0.15986667076746622
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,2,1,balanced,0.15891200304031372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,2,1,balanced,0.16683733463287354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,2,1,balanced,13.030794779459635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,2,1,balanced,0.2606079975763957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,2,power_law_1.2,1.0370816230773925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,2,1,balanced,0.21730132897694907
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,2,1,balanced,0.18756266434987387
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,2,1,balanced,0.4129013220469157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,2,1,balanced,0.2330133318901062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,2,1,balanced,0.3162826697031657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,2,1,balanced,0.4017866849899292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,2,1,balanced,0.575493335723877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,2,1,balanced,0.7485173543294271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,2,1,balanced,1.0977226893107097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,2,1,balanced,1.4506239891052246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,2,1,balanced,2.2220746676127114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,2,1,balanced,3.008085250854492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,2,1,power_law_1.01,0.13479679822921753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,2,1,power_law_1.01,0.16020480394363404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,2,1,power_law_1.01,0.2756160020828247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,2,1,balanced,17.106229146321613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,2,1,power_law_1.01,0.3656831979751587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,2,1,power_law_1.01,0.08064000010490417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,2,1,power_law_1.01,0.36494719982147217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,2,1,power_law_1.01,0.12307200431823731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,2,1,power_law_1.01,0.1552191972732544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,2,1,power_law_1.01,0.39567999839782714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,2,1,power_law_1.01,0.22636799812316893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,2,1,power_law_1.01,0.4287168025970459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,2,1,power_law_1.01,0.2264256000518799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,2,1,power_law_1.01,0.4566976070404053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,2,1,power_law_1.01,0.24467198848724364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,2,1,power_law_1.01,0.48114562034606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,2,1,power_law_1.01,0.2737407922744751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,2,1,power_law_1.01,0.30201599597930906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,2,1,power_law_1.01,0.49781122207641604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,2,1,power_law_1.01,0.298304009437561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,2,1,power_law_1.01,0.4336832046508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,2,1,balanced,0.26925865809122723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,2,1,power_law_1.01,0.3192768096923828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,2,1,balanced,0.26818132400512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,2,1,power_law_1.01,0.5056704044342041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,2,1,power_law_1.01,0.23971199989318848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,2,1,power_law_1.01,0.5160255908966065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,2,1,power_law_1.01,0.2842816114425659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,2,1,power_law_1.01,0.5429567813873291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,2,1,power_law_1.01,0.2902400016784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,2,1,power_law_1.01,0.3170815944671631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,2,1,power_law_1.01,0.6035711765289307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,2,1,power_law_1.01,0.34289278984069826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,2,1,power_law_1.01,0.6234752178192139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,2,1,power_law_1.01,0.35107200145721434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,2,1,power_law_1.01,0.42081918716430666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,2,1,power_law_1.01,0.7604159832000732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,2,1,power_law_1.01,0.4774208068847656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,2,1,power_law_1.01,0.8535039901733399
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,2,power_law_1.2,1.094752025604248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,2,1,power_law_1.01,0.5884992122650147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,2,1,power_law_1.01,1.05600004196167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,2,1,power_law_1.01,1.2600640296936034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,2,1,power_law_1.01,0.7048255920410156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,2,1,power_law_1.01,1.5555071830749512
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,2,power_law_1.2,1.5712703704833983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,2,1,power_law_1.01,0.8608832359313965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,2,power_law_1.2,14.303398132324219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,2,1,balanced,0.27322665850321454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,2,1,balanced,0.2813919981320699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,2,1,power_law_1.01,2.187443161010742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,2,power_law_1.2,7.842329406738282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,2,power_law_1.2,12.11310043334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,2,1,power_law_1.01,1.209164810180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,2,1,power_law_1.01,1.5629631996154785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,2,1,power_law_1.01,2.825926399230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,2,1,balanced,0.24186132351557413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,2,1,power_law_1.01,4.054528045654297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,2,1,power_law_1.01,2.2361215591430663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,2,1,power_law_1.01,2.9140287399291993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,2,power_law_1.2,20.38773193359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,2,1,balanced,0.2463573416074117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,2,1,balanced,0.25194666783014935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,2,1,balanced,0.25697600841522217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,2,1,balanced,0.338864008585612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,2,1,balanced,0.3460906744003296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,2,1,balanced,0.34573864936828613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,2,power_law_1.01,25.51564178466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,2,1,power_law_1.01,0.1065343976020813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,2,1,power_law_1.01,0.14111360311508178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,2,1,power_law_1.01,0.1679744005203247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,2,1,power_law_1.01,0.23208320140838623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,2,1,power_law_1.01,0.23104639053344728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,2,1,power_law_1.01,0.2457727909088135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,2,1,power_law_1.01,0.2619839906692505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,2,1,power_law_1.01,0.28462719917297363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,2,1,power_law_1.01,0.41448321342468264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,2,power_law_1.2,2.3294656753540037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,2,1,power_law_1.01,0.426636791229248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,2,1,power_law_1.01,0.44977917671203616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,2,1,power_law_1.01,0.4758143901824951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,2,1,balanced,0.4500266710917155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,2,1,power_law_1.01,0.4898367881774902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,2,1,balanced,0.4739146629969279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,2,1,power_law_1.01,0.5799808025360107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,2,1,power_law_1.01,5.567891311645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,2,1,power_law_1.01,0.6623231887817382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,2,1,power_law_1.01,0.7025216102600098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,2,1,power_law_1.01,0.849516773223877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,2,1,power_law_1.01,1.1244864463806152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,2,1,power_law_1.01,1.3756928443908691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,2,1,power_law_1.01,1.9278272628784179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,2,1,power_law_1.01,4.450451278686524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,2,1,power_law_1.01,2.4874048233032227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,2,1,balanced,0.6431253353754679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,2,1,balanced,0.803056001663208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,2,1,balanced,1.1457706292470295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,2,1,power_law_1.01,8.584153747558593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,2,1,balanced,1.4611040751139324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,2,1,balanced,2.220128059387207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,2,1,balanced,2.8866825103759766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,2,1,balanced,4.440581321716309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,2,power_law_1.2,19.118328857421876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,2,1,balanced,6.000629425048828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,2,1,power_law_1.2,0.14355839490890504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,2,1,power_law_1.2,0.1857151985168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,2,1,power_law_1.2,0.2684351921081543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,2,1,power_law_1.01,3.5692417144775392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,2,1,power_law_1.2,0.37240960597991946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,2,1,power_law_1.2,0.37930879592895506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,2,1,power_law_1.2,0.410368013381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,2,1,power_law_1.2,0.44376959800720217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,2,1,power_law_1.2,0.45582079887390137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,2,1,power_law_1.2,0.4776895999908447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,2,1,power_law_1.2,0.4942016124725342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,2,1,power_law_1.2,0.44538240432739257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,2,1,power_law_1.2,0.4935296058654785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,2,1,power_law_1.2,0.5277440071105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,2,1,power_law_1.2,0.55829758644104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,2,1,power_law_1.2,0.6041279792785644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,2,1,power_law_1.2,0.6355584144592286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,2,1,power_law_1.2,0.7700928211212158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,2,1,power_law_1.2,0.8350399971008301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,2,1,power_law_1.2,1.0589056015014648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,2,power_law_1.2,10.159033966064452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,2,1,power_law_1.2,0.09171839952468872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,2,1,power_law_1.2,0.1337280035018921
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,2,1,power_law_1.01,0.05315200090408325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,2,1,power_law_1.2,0.165011203289032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,2,1,power_law_1.01,0.06760960221290588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,2,1,power_law_1.2,0.2146239995956421
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,2,1,power_law_1.01,0.08919039964675904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,2,1,power_law_1.2,0.23758080005645751
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,2,1,power_law_1.01,0.12453759908676147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,2,1,power_law_1.2,0.2456831932067871
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,2,1,power_law_1.01,0.1232192039489746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,2,1,power_law_1.2,0.2769984006881714
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,2,1,power_law_1.01,0.13619199991226197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,2,1,power_law_1.2,0.2805567979812622
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,2,1,power_law_1.01,0.15059839487075805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,2,1,power_law_1.2,0.2946176052093506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,2,1,power_law_1.01,0.16997120380401612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,2,1,power_law_1.2,0.3243711948394775
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,2,1,power_law_1.01,0.1527616024017334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,2,1,power_law_1.2,0.24557440280914306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,2,1,power_law_1.01,0.156985604763031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,2,1,power_law_1.01,0.17260799407958985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,2,1,power_law_1.2,0.29103360176086424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,2,1,power_law_1.2,0.2960576057434082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,2,1,power_law_1.01,0.16168320178985596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,2,1,power_law_1.2,0.3114559888839722
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,2,1,power_law_1.01,0.17261439561843872
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,2,1,power_law_1.01,0.19631999731063843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,2,1,power_law_1.2,0.33827838897705076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,2,1,power_law_1.01,4.582815933227539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,2,1,power_law_1.2,1.235001564025879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,2,1,power_law_1.01,0.1915071964263916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,2,1,power_law_1.01,0.20273919105529786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,2,1,power_law_1.2,0.355513596534729
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,2,1,power_law_1.01,0.2316351890563965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,2,1,power_law_1.01,0.2928384065628052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,2,1,power_law_1.2,0.4169151782989502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,2,1,power_law_1.01,6.735174560546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,2,1,power_law_1.01,0.2811392068862915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,2,1,power_law_1.2,0.4803647994995117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,2,1,power_law_1.01,11.77175064086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,2,1,power_law_1.01,0.3766016006469727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,2,1,power_law_1.2,0.5923136234283447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,2,1,power_law_1.2,1.5659968376159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,2,1,power_law_1.01,0.4810431957244873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,2,1,power_law_1.01,6.147065734863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,2,1,power_law_1.01,0.6295551776885986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,2,1,power_law_1.2,2.1688575744628906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,2,1,power_law_1.2,0.7080575942993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,2,1,power_law_1.2,0.10505599975585937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,2,1,power_law_1.2,0.8792575836181641
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,2,1,power_law_1.01,0.7953728199005127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,2,1,power_law_1.2,0.12579200267791749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,2,1,power_law_1.2,0.13605120182037353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,2,1,power_law_1.01,1.1475263595581056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,2,1,power_law_1.2,0.05065600275993347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,2,1,power_law_1.2,0.0716480016708374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,2,1,power_law_1.2,0.0895359992980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,2,1,power_law_1.2,0.12062079906463623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,2,1,power_law_1.2,0.12505600452423096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,2,1,power_law_1.2,0.1331455945968628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,2,1,power_law_1.2,0.15259519815444947
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,2,1,power_law_1.2,0.16754560470581054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,2,1,power_law_1.2,0.15037440061569213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,2,1,power_law_1.2,0.15959680080413818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,2,1,power_law_1.2,0.17417600154876708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,2,1,power_law_1.2,0.16076799631118774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,2,1,power_law_1.2,0.17142399549484252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,2,1,power_law_1.2,0.20001280307769775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,2,1,power_law_1.2,0.22807679176330567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,2,1,power_law_1.2,0.23756799697875977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,2,1,power_law_1.2,2.7839040756225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,2,1,power_law_1.01,8.861459350585937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,2,power_law_1.2,4.019027328491211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,2,1,power_law_1.2,0.1923840045928955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,2,1,power_law_1.2,1.231827163696289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,2,1,power_law_1.2,0.20878078937530517
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,2,1,power_law_1.2,0.22500479221343994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,2,1,power_law_1.2,0.24661760330200194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,2,1,power_law_1.2,0.2454848051071167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,2,1,power_law_1.2,0.2614527940750122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,2,1,power_law_1.2,0.27243518829345703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,2,1,power_law_1.2,0.28677120208740237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,2,1,power_law_1.01,1.5070464134216308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,2,1,power_law_1.2,0.36970880031585696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,2,power_law_1.2,27.320050048828126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,2,1,power_law_1.2,1.5820480346679688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,2,1,power_law_1.2,0.4159872055053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,2,1,power_law_1.2,0.4348991870880127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,2,1,power_law_1.2,0.42247681617736815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,2,1,power_law_1.01,13.111672973632812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,2,1,power_law_1.2,0.4655871868133545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,2,1,power_law_1.2,0.47644801139831544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,2,1,power_law_1.2,0.5074304103851318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,2,balanced,0.08880533774693807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,2,balanced,0.15668267011642456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,2,balanced,0.24943999449412027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,2,1,power_law_1.2,4.242214584350586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,2,1,power_law_1.2,0.5846975803375244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,2,1,power_law_1.2,0.6384255886077881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,2,1,power_law_1.2,0.7194496154785156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,2,1,power_law_1.2,0.8560000419616699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,2,balanced,0.4341119925181071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,2,balanced,0.8339359760284424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,2,balanced,1.608239968617757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,2,balanced,1.6566294034322102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,2,balanced,1.6697866121927898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,2,balanced,1.8745919863382976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,2,balanced,1.874735991160075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,2,1,power_law_1.2,1.0988991737365723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,2,balanced,1.8697013854980469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,2,balanced,1.857103983561198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,2,balanced,1.8928693135579426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,2,balanced,1.902448018391927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,2,1,power_law_1.2,1.375385570526123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,2,balanced,1.9136053721110027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,2,balanced,1.9194614092508953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,2,1,power_law_1.2,0.6315392017364502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,2,balanced,1.9580480257670085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,2,balanced,2.28985595703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,2,balanced,2.3281866709391275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,2,balanced,2.419994672139486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,2,balanced,2.422335942586263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,2,1,power_law_1.2,1.931782341003418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,2,balanced,2.626405398050944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,2,balanced,2.8932746251424155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,2,1,power_law_1.01,2.1718399047851564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,2,balanced,4.49673589070638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,2,balanced,4.979370752970378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,2,1,power_law_1.2,2.448044776916504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,2,balanced,7.216890970865886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,2,1,power_law_1.2,5.578905487060547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,2,1,power_law_1.2,0.797056007385254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,2,balanced,9.195162455240885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,2,1,power_law_1.2,2.3022079467773438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,2,1,power_law_1.2,3.530790328979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,2,balanced,0.06759466727574666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,2,balanced,0.09328533212343852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,2,balanced,0.1523413360118866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,2,balanced,0.2632746696472168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,2,balanced,0.48161065578460693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,2,balanced,0.9241813023885092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,2,balanced,1.0368373394012451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,2,balanced,1.0432426929473877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,2,balanced,0.9893706639607748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,2,1,power_law_1.2,1.1343680381774903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,2,balanced,0.9924426873524984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,2,balanced,0.9705706437428793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,2,balanced,0.9835946559906006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,2,balanced,0.992415984471639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,2,1,power_law_1.2,4.585235214233398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,2,balanced,1.0059999624888103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,2,balanced,1.0179999669392903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,2,balanced,1.036842664082845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,2,1,power_law_1.01,17.363058471679686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,2,balanced,1.058789332707723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,2,balanced,1.2420799732208252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,2,balanced,1.275871992111206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,2,balanced,1.3486347198486328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,2,balanced,1.4287039438883464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,2,balanced,1.5891040166219075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,2,balanced,1.7536640167236328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,2,power_law_1.2,5.304915237426758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,2,balanced,2.6725918451944985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,2,balanced,3.087205251057943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,2,balanced,4.383978525797526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,2,balanced,5.809328079223633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,2,power_law_1.01,0.16172159910202027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,2,power_law_1.01,0.25004799365997316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,2,balanced,0.0807360013326009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,2,1,power_law_1.2,2.958451271057129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,2,balanced,0.1072746713956197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,2,balanced,0.15958933035532633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,2,balanced,0.2765760024388631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,2,balanced,0.5070453484853109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,2,balanced,0.9404853185017904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,2,balanced,0.9440426826477051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,2,balanced,0.9504426320393881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,2,balanced,0.03748266647259394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,2,balanced,0.05635199944178263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,2,balanced,0.08681600292523702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,2,balanced,0.14320000012715658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,2,power_law_1.01,0.3473727941513062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,2,balanced,0.25474133094151813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,2,balanced,0.46426133314768475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,2,power_law_1.01,0.5553152084350585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,2,balanced,0.46935466925303143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,2,1,power_law_1.2,1.4844927787780762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,2,balanced,0.47343464692433673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,2,balanced,0.4776000181833903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,2,power_law_1.01,0.7124864101409912
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,2,balanced,0.4800479809443156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,2,balanced,0.48677865664164227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,2,balanced,0.49431999524434406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,2,power_law_1.01,0.9441023826599121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,2,balanced,0.5041706562042236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,2,balanced,1.5392319361368816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,2,balanced,0.5170186758041382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,2,balanced,0.5074400107065836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,2,power_law_1.01,1.257932758331299
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,2,balanced,0.5180853207906088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,2,balanced,0.5357919931411743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,2,balanced,0.5456746816635132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,2,power_law_1.01,1.253171157836914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,2,balanced,0.5681120157241821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,2,balanced,0.5948319832483927
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,2,balanced,0.6253439982732137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,2,power_law_1.01,1.468825626373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,2,balanced,0.7861973444620768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,2,balanced,0.7440853118896484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,2,power_law_1.01,1.5379072189331056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,2,balanced,1.5466666221618652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,2,balanced,1.3154773712158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,2,balanced,1.5580053329467773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,2,balanced,1.229360024134318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,2,power_law_1.01,1.4630975723266602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,2,balanced,1.567962646484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,2,balanced,1.7458666165669758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,2,power_law_1.01,1.5604031562805176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,2,balanced,2.2962187131245932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,2,power_law_1.01,1.6208831787109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,2,power_law_1.01,1.6436223983764648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,2,1,power_law_1.01,2.9681215286254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,2,balanced,1.5802879333496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,2,power_law_1.01,1.6151487350463867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,2,balanced,1.6010613441467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,2,power_law_1.01,1.7378047943115233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,2,balanced,1.622165362040202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,2,balanced,1.6405493418375652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,2,power_law_1.01,1.7396095275878907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,2,balanced,1.6746986707051594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,2,balanced,1.7403039932250977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,2,power_law_1.01,2.053004837036133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,2,balanced,1.7977066040039062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,2,power_law_1.01,2.1569536209106444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,2,balanced,1.915269374847412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,2,power_law_1.01,0.08807039856910706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,2,power_law_1.01,0.14865920543670655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,2,balanced,2.0388320287068686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,2,power_law_1.01,0.20382080078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,2,power_law_1.01,2.457088088989258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,2,power_law_1.01,0.29580159187316896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,2,balanced,3.374095916748047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,2,power_law_1.01,2.5747968673706056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,2,1,power_law_1.2,4.4449409484863285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,2,balanced,3.6077868143717446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,2,power_law_1.01,3.1709760665893554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,2,power_law_1.01,0.4328256130218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,2,balanced,5.226976076761882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,2,power_law_1.01,0.49338879585266116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,2,power_law_1.01,0.7559040069580079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,2,power_law_1.01,3.59947509765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,2,power_law_1.01,0.10372480154037475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,2,power_law_1.01,0.8261631965637207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,2,power_law_1.01,0.15525120496749878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,2,power_law_1.01,0.2436288118362427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,2,power_law_1.01,0.7826560020446778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,2,balanced,6.8220640818278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,2,power_law_1.01,0.3654592037200928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,2,power_law_1.01,0.8403712272644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,2,power_law_1.01,0.44832639694213866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,2,power_law_1.01,4.785145568847656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,2,power_law_1.01,0.8268671989440918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,2,power_law_1.01,0.5856768131256104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,2,power_law_1.01,0.8594367980957032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,2,1,power_law_1.2,8.653958129882813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,2,power_law_1.01,0.7277120113372803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,2,power_law_1.01,0.8571840286254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,2,power_law_1.01,0.771340799331665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,2,power_law_1.01,0.8886336326599121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,2,balanced,10.02011744181315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,2,power_law_1.01,5.917049789428711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,2,power_law_1.01,0.9260671615600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,2,power_law_1.01,1.2123135566711425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,2,1,power_law_1.2,6.722297668457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,2,power_law_1.01,0.929088020324707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,2,power_law_1.01,1.3504128456115723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,2,power_law_1.01,0.9911552429199219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,2,power_law_1.01,1.3340736389160157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,2,power_law_1.01,1.1666687965393066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,2,power_law_1.01,1.3500032424926758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,2,power_law_1.01,1.2587264060974122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,2,power_law_1.01,7.803033447265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,2,power_law_1.01,1.3971648216247559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,2,power_law_1.01,1.3914112091064452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,2,balanced,13.176512400309244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,2,power_law_1.01,1.4106304168701171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,2,power_law_1.01,1.552614402770996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,2,power_law_1.01,1.4701631546020508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,2,power_law_1.01,1.8480447769165038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,2,power_law_1.01,1.5652223587036134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,2,power_law_1.01,2.1744512557983398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,2,power_law_1.01,1.5788800239562988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,2,power_law_1.01,10.708038330078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,2,power_law_1.01,1.8089792251586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,2,power_law_1.01,2.7582975387573243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,2,power_law_1.2,0.17240320444107055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,2,power_law_1.01,2.003046417236328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,2,power_law_1.2,0.27733120918273924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,2,power_law_1.2,0.33854720592498777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,2,power_law_1.01,3.5032638549804687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,2,power_law_1.2,0.5245952129364013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,2,power_law_1.01,2.400595283508301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,2,power_law_1.2,0.7208000183105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,2,power_law_1.2,0.805299186706543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,2,power_law_1.01,2.8328319549560548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,2,power_law_1.01,4.842623901367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,2,power_law_1.2,1.2742400169372559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,2,power_law_1.2,1.221388816833496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,2,power_law_1.01,3.710406494140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,2,power_law_1.01,0.05722879767417908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,2,power_law_1.2,1.4269248008728028
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,2,power_law_1.01,0.08364800214767457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,2,power_law_1.01,6.0485694885253904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,2,power_law_1.01,0.11061760187149047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,2,power_law_1.2,1.5599488258361816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,2,power_law_1.01,4.662886428833008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,2,power_law_1.01,0.16402560472488403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,2,power_law_1.2,1.45665283203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,2,power_law_1.01,0.23016319274902344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,2,power_law_1.2,0.08983680009841918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,2,power_law_1.2,0.14956159591674806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,2,power_law_1.01,0.28627839088439944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,2,power_law_1.2,0.2019263982772827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,2,power_law_1.2,1.5746560096740723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,2,power_law_1.01,0.36294400691986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,2,power_law_1.2,0.288703989982605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,2,power_law_1.01,0.38165760040283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,2,power_law_1.2,1.6068607330322267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,2,power_law_1.2,0.410368013381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,2,power_law_1.01,0.4051328182220459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,2,power_law_1.01,6.269113540649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,2,power_law_1.2,0.5305024147033691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,2,power_law_1.01,0.40280961990356445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,2,power_law_1.2,1.6023168563842773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,2,power_law_1.2,0.7806848049163818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,2,power_law_1.01,0.4246848106384277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,2,power_law_1.2,1.678438377380371
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,2,power_law_1.01,0.4368127822875977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,2,power_law_1.2,0.8189824104309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,2,power_law_1.01,0.466099214553833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,2,power_law_1.2,0.856595230102539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,2,power_law_1.2,1.6858240127563477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,2,power_law_1.01,0.5067200183868408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,2,power_law_1.2,0.8461376190185547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,2,power_law_1.01,0.45810561180114745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,2,power_law_1.2,1.777440071105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,2,power_law_1.01,7.9889984130859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,2,power_law_1.2,0.8121536254882813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,2,power_law_1.01,0.4939136028289795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,2,power_law_1.01,0.537446403503418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,2,power_law_1.2,0.8438464164733886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,2,power_law_1.2,2.162585639953613
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,2,power_law_1.01,0.5438015937805176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,2,power_law_1.2,0.8950271606445312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,2,power_law_1.01,0.6099008083343506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,2,power_law_1.2,0.9040575981140136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,2,power_law_1.2,2.31213436126709
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,2,power_law_1.01,0.6358528137207031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,2,power_law_1.2,0.9314111709594727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,2,power_law_1.01,0.7275328159332275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,2,power_law_1.2,0.961734390258789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,2,power_law_1.2,2.590412712097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,2,power_law_1.2,0.999392032623291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,2,power_law_1.2,1.2346176147460937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,2,power_law_1.01,11.110054779052735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,2,power_law_1.2,2.798316764831543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,2,power_law_1.2,1.2921216011047363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,2,1,power_law_1.2,2.1740352630615236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,2,power_law_1.01,0.8471487998962403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,2,power_law_1.2,1.4343487739562988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,2,power_law_1.2,3.3673023223876952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,2,power_law_1.01,0.9762175559997559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,2,power_law_1.2,1.6478271484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,2,power_law_1.01,1.2248448371887206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,2,power_law_1.2,1.9288959503173828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,2,power_law_1.2,3.8399295806884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,2,power_law_1.01,1.536838436126709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,2,power_law_1.2,2.324224090576172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,2,power_law_1.01,2.108255958557129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,2,power_law_1.2,2.919571113586426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,2,power_law_1.01,3.5895614624023438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,2,power_law_1.01,14.54676513671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,2,power_law_1.2,3.6107521057128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,2,power_law_1.2,5.11357421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,2,power_law_1.2,5.1177215576171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,2,power_law_1.2,0.0996288001537323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,2,1,power_law_1.2,8.81844482421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,2,power_law_1.2,0.15486719608306884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,2,power_law_1.2,0.2425856113433838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,2,power_law_1.2,0.31086719036102295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,2,power_law_1.2,0.42771201133728026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,2,power_law_1.2,6.068057632446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,2,power_law_1.2,0.5147391796112061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,2,power_law_1.2,6.5709693908691404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,2,1,balanced,0.08069866895675659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,2,power_law_1.2,0.7538623809814453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,2,1,balanced,0.15503467122713724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,2,1,balanced,0.2526666720708211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,2,1,balanced,0.4392533302307129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,2,power_law_1.2,0.7085631847381592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,2,1,balanced,0.8494079907735189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,2,1,balanced,1.6467572848002117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,2,power_law_1.2,1.2590911865234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,2,1,balanced,1.7123200098673503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,2,power_law_1.2,1.2806976318359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,2,1,balanced,1.715269406636556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,2,power_law_1.2,8.603660583496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,2,power_law_1.2,1.2906240463256835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,2,power_law_1.2,0.05699840188026428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,2,1,balanced,1.8799519538879395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,2,power_law_1.2,0.08294399976730346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,2,1,balanced,1.8775092760721843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,2,power_law_1.2,1.344940757751465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,2,power_law_1.2,0.11062400341033936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,2,power_law_1.2,0.1615872025489807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,2,1,balanced,1.8417599995930989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,2,power_law_1.2,1.4240575790405274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,2,power_law_1.2,0.20626559257507324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,2,1,balanced,1.839135964711507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,2,power_law_1.2,0.26344320774078367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,2,power_law_1.2,1.4311039924621582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,2,1,balanced,1.8851253191630046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,2,power_law_1.2,0.3782912015914917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,2,power_law_1.2,0.35110399723052976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,2,power_law_1.2,1.5307007789611817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,2,power_law_1.2,0.38974719047546386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,2,power_law_1.2,1.5005887985229491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,2,power_law_1.2,10.972505950927735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,2,power_law_1.2,0.3922368049621582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,2,1,power_law_1.2,6.109503936767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,2,power_law_1.2,0.4346367835998535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,2,1,balanced,0.0693333347638448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,2,1,balanced,0.09732266267140706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,2,power_law_1.2,1.6635391235351562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,2,1,balanced,0.15949333707491556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,2,power_law_1.2,0.43450241088867186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,2,1,balanced,0.27582399050394696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,2,1,balanced,0.49790934721628827
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,2,power_law_1.2,0.4819327831268311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,2,1,balanced,1.8650347391764324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,2,power_law_1.2,1.8895360946655273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,2,power_law_1.2,0.5111551761627198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,2,1,balanced,1.829103946685791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,2,power_law_1.2,0.47533440589904785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,2,power_law_1.2,0.5070847988128662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,2,1,balanced,1.8357067108154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,2,power_law_1.2,2.0741695404052733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,2,power_law_1.2,0.5721536159515381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,2,1,balanced,1.881648063659668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,2,power_law_1.2,0.5779647827148438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,2,1,balanced,0.07297066847483318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,2,1,balanced,0.09623466928799947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,2,1,balanced,0.9416266282399496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,2,1,power_law_1.2,11.599763488769531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,2,1,balanced,0.14858133594195047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,2,power_law_1.2,2.4893951416015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,2,1,balanced,2.1866453488667807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,2,power_law_1.2,0.6586431980133056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,2,1,balanced,1.0628320376078289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,2,1,balanced,1.0655999978383381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,2,power_law_1.2,0.6545792102813721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,2,1,balanced,2.2208639780680337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,2,1,balanced,0.993717352549235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,2,power_law_1.2,0.7733503818511963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,2,power_law_1.2,2.9889984130859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,2,1,balanced,2.2911574045817056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,2,power_law_1.2,0.9251263618469239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,2,1,balanced,2.3540053367614746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,2,1,balanced,0.2608213424682617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,2,1,balanced,0.4989866813023885
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,2,power_law_1.2,1.0764863967895508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,2,1,balanced,0.9458773136138916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,2,power_law_1.2,3.906982421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,2,1,balanced,2.5830933252970376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,2,1,balanced,0.9520426591237386
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,2,power_law_1.2,1.338931179046631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,2,1,balanced,0.9951146443684896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,2,1,balanced,0.9540639718373617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,2,1,balanced,2.678256034851074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,2,1,balanced,1.561365286509196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,2,power_law_1.2,1.6926464080810546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,2,1,balanced,1.568895975748698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,2,power_law_1.2,4.524921417236328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,2,1,balanced,4.18504524230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,2,power_law_1.2,2.181222343444824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,2,1,balanced,0.93777068456014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,2,1,balanced,4.489525477091472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,2,power_law_1.2,4.031334304809571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,2,power_law_1.2,6.219488143920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,2,1,balanced,1.577664057413737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,2,1,power_law_1.01,0.07550079822540283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,2,1,balanced,1.5834879875183105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,2,1,power_law_1.01,0.14416639804840087
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,2,1,balanced,0.037733333806196846
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,2,1,balanced,0.058559998869895935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,2,1,balanced,0.08720533053080241
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,2,1,balanced,0.1411893367767334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,2,1,balanced,6.914938608805339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,2,1,balanced,1.592458724975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,2,1,balanced,0.24970666567484537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,2,1,balanced,0.9448906580607096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,2,1,balanced,0.46530667940775555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,2,1,balanced,1.6081013679504395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,2,1,balanced,0.47119998931884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,2,1,balanced,0.9505440394083658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,2,1,balanced,0.47667733828226727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,2,1,balanced,0.9576480388641357
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,2,1,balanced,0.47997868061065674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,2,1,balanced,1.6207946141560872
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,2,1,balanced,0.4824320077896118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,2,1,balanced,0.9691946506500244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,2,1,balanced,0.4875359932581584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,2,1,balanced,0.9752693176269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,2,1,balanced,1.6364693641662598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,2,1,balanced,0.4946719805399577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,2,1,balanced,0.5021386543909708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,2,1,balanced,0.9941066900889078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,2,1,balanced,0.5144960085550944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,2,1,balanced,1.6584320068359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,2,1,power_law_1.01,0.24274559020996095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,2,1,balanced,0.5488959948221842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,2,1,balanced,1.2321226596832275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,2,1,balanced,8.679951985677084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,2,1,balanced,0.5591093301773071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,2,1,balanced,1.701877276102702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,2,1,balanced,1.2568853696187336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,2,1,balanced,0.585536003112793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,2,1,balanced,0.592298666636149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,2,1,balanced,1.2940800189971924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,2,1,balanced,1.7367733319600422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,2,1,balanced,0.6223786671956381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,2,1,power_law_1.01,0.06815999746322632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,2,1,balanced,0.6827733516693115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,2,1,balanced,1.3486026128133137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,2,1,power_law_1.01,0.10083199739456176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,2,1,balanced,1.8155466715494792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,2,1,balanced,0.7327573299407959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,2,1,power_law_1.01,0.14540799856185913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,2,1,balanced,1.4647146860758464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,2,power_law_1.2,8.094182586669922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,2,1,balanced,0.8993173440297445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,2,1,power_law_1.01,0.25340800285339354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,2,1,balanced,1.902880032857259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,2,1,balanced,1.593920071919759
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,2,1,balanced,0.9411253134409586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,2,1,power_law_1.01,0.4301439762115479
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,2,1,balanced,1.7387839953104656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,2,1,balanced,2.5043039321899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,2,1,power_law_1.01,0.6564032077789307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,2,1,balanced,3.22981325785319
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,2,1,balanced,1.326416015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,2,1,power_law_1.01,0.8918720245361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,2,1,balanced,2.7689491907755532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,2,1,balanced,1.8751145998636882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,2,1,balanced,3.395397186279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,2,1,power_law_1.01,1.2218815803527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,2,1,power_law_1.01,0.36452479362487794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,2,1,balanced,2.464944044748942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,2,1,balanced,4.015184084574382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,2,1,power_law_1.01,1.2914048194885255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,2,1,power_law_1.01,0.48152961730957033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,2,1,power_law_1.01,0.7866367816925048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,2,1,balanced,4.929189364115397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,2,power_law_1.2,11.027839660644531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,2,1,power_law_1.01,1.4568448066711426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,2,1,power_law_1.01,0.8107904434204102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,2,1,balanced,5.149264017740886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,2,1,power_law_1.01,1.5005696296691895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,2,1,power_law_1.01,0.7792319774627685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,2,1,power_law_1.01,1.4928768157958985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,2,1,balanced,6.4301706949869795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,2,1,power_law_1.01,1.5104384422302246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,2,1,power_law_1.01,0.07192320227622986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,2,1,power_law_1.01,0.09472640156745911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,2,1,power_law_1.01,1.531494426727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,2,1,power_law_1.01,0.1472767949104309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,2,1,power_law_1.01,0.2610368013381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,2,1,power_law_1.01,0.8036735534667969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,2,1,power_law_1.01,1.608083152770996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,2,1,power_law_1.01,0.3921087980270386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,2,1,power_law_1.01,0.7689151763916016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,2,1,balanced,9.432629267374674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,2,1,power_law_1.01,1.6045696258544921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,2,1,power_law_1.01,0.5164224147796631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,2,1,power_law_1.01,0.8026687622070312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,2,1,power_law_1.01,0.681171178817749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,2,1,power_law_1.01,0.8098048210144043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,2,1,power_law_1.01,1.6665664672851563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,2,power_law_1.2,14.883705139160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,2,1,power_law_1.01,0.7312511920928955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,2,1,power_law_1.01,0.8564160346984864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,2,1,power_law_1.2,0.07564799785614014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,2,1,power_law_1.01,1.7056640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,2,1,power_law_1.01,0.8395648002624512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,2,1,power_law_1.2,0.14531840085983277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,2,1,power_law_1.2,13.04901123046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,2,1,power_law_1.2,0.24238080978393556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,2,1,power_law_1.01,0.8773695945739746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,2,1,power_law_1.01,2.025984001159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,2,1,power_law_1.2,0.4222080230712891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,2,1,power_law_1.01,0.9215935707092285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,2,1,power_law_1.2,0.5914112091064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,2,1,power_law_1.01,1.147379207611084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,2,1,power_law_1.01,2.162899208068848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,2,1,power_law_1.2,0.804684829711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,2,1,power_law_1.01,1.2788607597351074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,2,1,power_law_1.01,1.2163904190063477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,2,1,power_law_1.01,0.03776639997959137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,2,1,power_law_1.2,1.2127615928649902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,2,1,power_law_1.01,1.296678352355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,2,1,power_law_1.01,0.0577023983001709
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,2,1,power_law_1.01,0.08480640053749085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,2,1,power_law_1.2,1.2715264320373536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,2,1,power_law_1.01,1.3530176162719727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,2,1,power_law_1.01,0.13826559782028197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,2,1,balanced,12.418266296386719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,2,1,power_law_1.01,0.20163838863372802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,2,1,power_law_1.01,1.3400256156921386
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,2,1,power_law_1.01,0.26372480392456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,2,1,power_law_1.2,0.05550079941749573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,2,1,power_law_1.01,2.3320255279541016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,2,1,power_law_1.01,0.3588736057281494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,2,1,power_law_1.2,0.08144639730453491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,2,1,power_law_1.01,1.4031935691833497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,2,1,power_law_1.01,1.3505599975585938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,2,1,power_law_1.2,0.14287999868392945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,2,1,power_law_1.01,0.3763391971588135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,2,1,power_law_1.2,0.23767681121826173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,2,1,power_law_1.01,1.455635166168213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,2,1,power_law_1.01,1.4699135780334474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,2,1,power_law_1.01,0.3824512004852295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,2,1,power_law_1.2,0.33489279747009276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,2,1,power_law_1.2,1.4379712104797364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,2,1,power_law_1.01,0.39604480266571046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,2,1,power_law_1.01,1.417523193359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,2,1,power_law_1.2,0.43686399459838865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,2,1,power_law_1.01,1.7459072113037108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,2,1,power_law_1.01,0.40615038871765136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,2,1,power_law_1.2,1.4817536354064942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,2,1,power_law_1.01,0.4279935836791992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,2,1,power_law_1.2,0.758464002609253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,2,1,power_law_1.01,1.48602237701416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,2,1,power_law_1.01,1.99552001953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,2,1,power_law_1.01,0.4433023929595947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,2,1,power_law_1.2,0.8075136184692383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,2,1,power_law_1.2,1.4724415779113769
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,2,1,power_law_1.01,0.46898560523986815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,2,1,power_law_1.01,1.6057024002075195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,2,1,power_law_1.01,2.4837440490722655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,2,1,power_law_1.2,0.7916351795196533
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,2,1,power_law_1.01,0.5117311954498291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,2,1,power_law_1.01,2.5792640686035155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,2,1,power_law_1.2,0.8083711624145508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,2,1,power_law_1.01,0.5547776222229004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,2,1,power_law_1.01,1.7815616607666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,2,1,power_law_1.01,0.640505599975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,2,1,power_law_1.01,2.943712043762207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,2,1,power_law_1.01,0.6139008045196533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,2,1,power_law_1.01,1.967308807373047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,2,1,power_law_1.01,3.195347213745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,2,1,power_law_1.01,0.7080704212188721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,2,1,power_law_1.01,3.3876991271972656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,2,1,power_law_1.01,0.7231743812561036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,2,1,power_law_1.2,1.5436800003051758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,2,1,power_law_1.01,0.8454912185668946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,2,1,power_law_1.2,0.7769536018371582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,2,1,power_law_1.2,1.5499967575073241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,2,1,power_law_1.01,4.265836715698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,2,1,power_law_1.01,0.9000703811645507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,2,1,power_law_1.2,0.798854398727417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,2,1,power_law_1.01,4.3927360534667965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,2,1,power_law_1.2,1.613145637512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,2,1,power_law_1.01,1.071123218536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,2,1,power_law_1.2,0.8262207984924317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,2,1,power_law_1.01,2.3274112701416017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,2,1,power_law_1.2,1.652511978149414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,2,1,power_law_1.01,1.2706111907958983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,2,1,power_law_1.01,5.351481628417969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,2,1,power_law_1.2,1.6967039108276367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,2,1,power_law_1.01,1.541971206665039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,2,1,power_law_1.01,2.6900608062744142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,2,1,power_law_1.01,5.118348693847656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,2,1,power_law_1.2,1.7619583129882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,2,1,power_law_1.01,2.093926429748535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,2,1,power_law_1.01,3.4297088623046874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,2,1,power_law_1.2,2.1767744064331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,2,1,power_law_1.01,3.13187198638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,2,1,power_law_1.2,2.2592063903808595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,2,1,power_law_1.01,7.344786834716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,2,1,power_law_1.2,0.07090560197830201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,2,1,power_law_1.01,4.150137710571289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,2,1,power_law_1.2,0.838809585571289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,2,1,power_law_1.2,2.411622428894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,2,1,power_law_1.2,0.8810560226440429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,2,1,power_law_1.2,0.8809984207153321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,2,1,power_law_1.2,0.9377344131469727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,2,1,power_law_1.2,2.7054399490356444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,2,1,power_law_1.01,5.647545623779297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,2,1,power_law_1.2,1.173740768432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,2,1,power_law_1.01,9.27352294921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,2,1,power_law_1.2,3.1586944580078127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,2,1,power_law_1.2,1.2720512390136718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,2,1,power_law_1.2,1.3913279533386231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,2,1,power_law_1.2,3.6242366790771485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,2,1,power_law_1.2,1.5198847770690918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,2,1,power_law_1.01,7.065574645996094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,2,1,power_law_1.2,1.8027519226074218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,2,1,power_law_1.2,0.0949567973613739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,2,1,power_law_1.2,4.64922866821289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,2,1,power_law_1.2,2.086579132080078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,2,1,power_law_1.2,0.03666560053825378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,2,balanced,0.07622399926185608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,2,balanced,0.13372266292572021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,2,1,power_law_1.2,2.646348762512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,2,1,power_law_1.2,0.0578495979309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,2,balanced,0.21945067246754965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,2,balanced,0.37756800651550293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,2,1,power_law_1.2,0.08546559810638428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,2,balanced,0.7184426784515381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,2,1,power_law_1.01,9.873516845703126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,2,1,power_law_1.2,5.5758720397949215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,2,1,power_law_1.2,0.1295807957649231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,2,1,power_law_1.2,0.14554879665374756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,2,balanced,1.3840373357137044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,2,1,power_law_1.2,0.24400639533996582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,2,1,power_law_1.2,0.18071680068969725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,2,balanced,1.4255520502726238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,2,1,power_law_1.2,3.2479488372802736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,2,1,power_law_1.2,0.3568959951400757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,2,1,power_law_1.2,0.2450176000595093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,2,balanced,1.4395999908447266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,2,1,power_law_1.2,0.484665584564209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,2,balanced,1.5897760391235352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,2,1,power_law_1.2,0.6817599773406983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,2,balanced,1.5945013364156086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,2,1,power_law_1.2,0.7126143932342529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,2,1,power_law_1.2,4.398457717895508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,2,balanced,1.5693386395772297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,2,1,power_law_1.2,7.514291381835937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,2,balanced,1.5709333419799805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,2,1,power_law_1.2,1.2426815986633302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,2,balanced,1.5696959495544434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,2,1,power_law_1.2,0.35630719661712645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,2,1,power_law_1.2,1.2747584342956544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,2,balanced,1.5812907218933105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,2,1,power_law_1.2,0.3732671976089478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,2,1,power_law_1.01,12.658649444580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,2,balanced,1.5941920280456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,2,1,power_law_1.2,1.3117631912231444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,2,1,power_law_1.2,5.581062316894531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,2,balanced,1.6570879618326824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,2,1,power_law_1.2,1.3651136398315429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,2,balanced,1.676037311553955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,2,1,power_law_1.2,2.917702484130859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,2,1,power_law_1.2,1.3945728302001954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,2,balanced,1.9207199414571126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,2,1,power_law_1.2,9.29217300415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,2,1,power_law_1.2,1.425216007232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,2,balanced,1.9546027183532715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,2,1,power_law_1.2,0.38664960861206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,2,1,power_law_1.2,0.38460800647735593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,2,balanced,2.0185759862264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,2,1,power_law_1.2,1.4929023742675782
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,2,1,power_law_1.2,0.41111040115356445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,2,balanced,2.1683680216471353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,2,1,power_law_1.2,0.4268799781799316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,2,1,power_law_1.2,1.553446388244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,2,1,power_law_1.2,0.444268798828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,2,balanced,2.301685333251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,2,1,power_law_1.2,0.48798718452453616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,2,1,power_law_1.2,1.674367904663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,2,1,power_law_1.2,0.5320320129394531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,2,balanced,0.06434133152167003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,2,balanced,0.08720533053080241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,2,balanced,2.475674629211426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,2,balanced,0.1411146620909373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,2,balanced,0.23292799790700278
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,2,1,power_law_1.2,0.5657536029815674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,2,balanced,0.4186720053354899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,2,1,power_law_1.2,1.8378496170043945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,2,balanced,0.795087973276774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,2,1,power_law_1.2,0.6695040225982666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,2,balanced,0.8978986740112305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,2,1,power_law_1.2,0.6343999862670898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,2,balanced,4.119898796081543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,2,balanced,0.8993493715922037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,2,1,power_law_1.2,2.0428672790527345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,2,balanced,0.8553760051727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,2,1,power_law_1.2,0.7363200187683105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,2,balanced,0.8597546418507894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,2,power_law_1.01,0.1299839973449707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,2,balanced,0.8462773164113363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,2,1,power_law_1.2,0.7511104106903076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,2,power_law_1.01,0.21683840751647948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,2,balanced,0.8611093362172445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,2,balanced,4.286954561869304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,2,1,power_law_1.2,2.3923007965087892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,2,balanced,0.8622026443481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,2,power_law_1.01,0.3017983913421631
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,2,1,power_law_1.2,0.8918208122253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,2,balanced,0.870474656422933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,2,power_law_1.01,0.4923520088195801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,2,balanced,0.8835840225219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,2,1,power_law_1.2,0.9371647834777832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,2,balanced,0.893509308497111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,2,1,power_law_1.2,2.79553279876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,2,power_law_1.01,0.6333312034606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,2,balanced,0.07622933387756348
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,2,balanced,0.03186666717131933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,2,balanced,0.09899733463923137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,2,balanced,0.9171093304951986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,2,balanced,0.05349866549173991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,2,balanced,0.14171733458836874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,2,balanced,0.07962666451931
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,2,1,power_law_1.2,1.1093759536743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,2,balanced,6.349045435587565
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,2,balanced,0.12760000427563986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,2,balanced,1.0771040121714275
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,2,balanced,0.22492265701293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,2,power_law_1.01,0.7949440002441406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,2,balanced,0.40380799770355225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,2,balanced,1.108837366104126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,2,balanced,0.40615467230478924
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,2,1,power_law_1.2,1.336032009124756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,2,balanced,0.4106666644414266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,2,power_law_1.01,1.1110207557678222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,2,1,power_law_1.2,3.5031806945800783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,2,balanced,1.1788427035013835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,2,balanced,0.414031982421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,2,balanced,0.41621867815653485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,2,balanced,1.2507039705912273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,2,power_law_1.01,1.1598719596862792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,2,1,power_law_1.2,1.5909248352050782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,2,balanced,1.3920693397521973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,2,balanced,0.23723200956980386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,2,balanced,7.9879201253255205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,2,balanced,1.5385120709737141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,2,balanced,0.44204266866048175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,2,1,power_law_1.2,4.241952133178711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,2,balanced,0.8086773554484049
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,2,1,power_law_1.2,2.1299840927124025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,2,balanced,0.8141760031382242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,2,balanced,2.3187573750813804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,2,balanced,0.8162453174591064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,2,balanced,0.42430933316548664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,2,balanced,1.3243093490600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,2,balanced,0.43055999279022217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,2,balanced,2.6432107289632163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,2,balanced,0.4383893410364787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,2,balanced,1.3305333455403645
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,2,balanced,0.4513653516769409
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,2,1,power_law_1.2,3.2024383544921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,2,balanced,0.4427093267440796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,2,power_law_1.01,1.3072447776794434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,2,balanced,1.3400586446126301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,2,balanced,0.4490773280461629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,2,1,power_law_1.2,5.681977462768555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,2,balanced,1.3514399528503418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,2,balanced,3.7464640935262046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,2,power_law_1.01,1.2966143608093261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,2,balanced,1.3606079419453938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,2,balanced,1.3820746739705403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,2,power_law_1.01,1.2606783866882325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,2,balanced,1.3999253908793132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,2,balanced,4.976016044616699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,2,balanced,1.41648530960083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,2,balanced,0.4660053253173828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,2,power_law_1.01,0.0802623987197876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,2,balanced,1.450170675913493
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,2,balanced,0.47173333168029785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,2,power_law_1.01,0.13589760065078735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,2,balanced,0.4935520092646281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,2,balanced,1.509658654530843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,2,balanced,0.5149759848912557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,2,power_law_1.01,0.1933632016181946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,2,balanced,0.5414880116780599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,2,balanced,1.562453269958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,2,power_law_1.01,0.2781248092651367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,2,balanced,0.7002240022023519
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,2,balanced,0.6411679983139038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,2,power_law_1.01,0.3662911891937256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,2,balanced,1.6715839703877766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,2,power_law_1.01,1.337504005432129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,2,balanced,1.1833066940307617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,2,power_law_1.01,0.47251200675964355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,2,balanced,1.7816267013549805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,2,balanced,1.063754638036092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,2,power_law_1.01,1.3598336219787597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,2,1,power_law_1.2,7.0857666015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,2,balanced,1.514069398244222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,2,balanced,2.9448960622151694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,2,power_law_1.01,1.3630080223083496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,2,balanced,1.9596266746520996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,2,power_law_1.01,1.3966079711914063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,2,balanced,3.157072067260742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,2,power_law_1.01,1.3879615783691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,2,power_law_1.01,0.6580607891082764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,2,power_law_1.01,1.5188223838806152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,2,power_law_1.01,0.6973184108734131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,2,balanced,4.571749369303386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,2,power_law_1.01,0.678764820098877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,2,1,power_law_1.2,17.314451599121092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,2,power_law_1.01,1.730739212036133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,2,1,power_law_1.2,9.787366485595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,2,power_law_1.01,0.706873607635498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,2,power_law_1.01,0.7364927768707276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,2,power_law_1.01,1.8216127395629882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,2,balanced,5.967226664225261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,2,power_law_1.01,0.7659776210784912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,2,power_law_1.01,2.0419647216796877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,2,power_law_1.01,0.7817920207977295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,2,power_law_1.01,0.774560022354126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,2,power_law_1.01,2.2567935943603517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,2,power_law_1.2,0.13240959644317626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,2,power_law_1.2,0.21794559955596923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,2,balanced,8.755274454752604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,2,power_law_1.2,0.28702080249786377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,2,power_law_1.01,2.671808052062988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,2,power_law_1.2,0.4507008075714111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,2,power_law_1.01,0.05259519815444946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,2,power_law_1.01,0.07622399926185608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,2,power_law_1.2,0.5787327766418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,2,1,power_law_1.2,12.51296615600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,2,power_law_1.01,0.10419199466705323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,2,power_law_1.01,0.7780992031097412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,2,power_law_1.2,0.6903168201446533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,2,power_law_1.01,3.13558406829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,2,power_law_1.01,0.15199359655380248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,2,power_law_1.01,0.8129023551940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,2,power_law_1.01,0.20436480045318603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,2,power_law_1.2,1.1040255546569824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,2,power_law_1.01,0.8550335884094238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,2,power_law_1.01,0.2497920036315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,2,power_law_1.2,1.136134433746338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,2,power_law_1.01,0.32908799648284914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,2,power_law_1.01,1.0211135864257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,2,balanced,11.516410827636719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,2,power_law_1.01,4.057945632934571
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,2,power_law_1.01,0.3329983949661255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,2,power_law_1.2,1.2939200401306152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,2,power_law_1.01,1.1128959655761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,2,power_law_1.01,0.32391040325164794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,2,power_law_1.01,0.09219200015068055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,2,power_law_1.2,1.3147007942199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,2,power_law_1.01,1.2138175964355469
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,2,power_law_1.01,0.35602560043334963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,2,power_law_1.01,0.14379520416259767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,2,power_law_1.01,0.36433279514312744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,2,power_law_1.01,0.22137598991394042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,2,power_law_1.01,1.3588288307189942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,2,power_law_1.2,1.3365376472473145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,2,power_law_1.01,0.38499200344085693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,2,power_law_1.01,5.237203216552734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,2,power_law_1.01,0.2910527944564819
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,2,power_law_1.01,0.40020480155944826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,2,power_law_1.01,0.40692481994628904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,2,power_law_1.2,1.258579158782959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,2,power_law_1.01,1.6326080322265626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,2,power_law_1.01,0.43089280128479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,2,power_law_1.01,0.4497920036315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,2,power_law_1.01,0.4064320087432861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,2,power_law_1.2,1.3783616065979003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,2,power_law_1.01,0.6376639842987061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,2,power_law_1.01,1.963199996948242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,2,power_law_1.01,0.41960959434509276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,2,power_law_1.2,1.4463232040405274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,2,power_law_1.01,0.6369855880737305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,2,power_law_1.01,1.097599983215332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,2,power_law_1.01,2.498956871032715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,2,power_law_1.01,7.49722900390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,2,power_law_1.01,1.0882047653198241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,2,power_law_1.2,0.07796480059623719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,2,power_law_1.01,1.1409215927124023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,2,power_law_1.01,3.0643007278442385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,2,power_law_1.2,0.13114880323410033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,2,power_law_1.01,0.459116792678833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,2,power_law_1.2,0.17407360076904296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,2,power_law_1.01,1.1720255851745605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,2,power_law_1.01,0.47084798812866213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,2,power_law_1.2,1.4287360191345215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,2,power_law_1.2,0.2560447931289673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,2,power_law_1.01,0.5268799781799316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,2,power_law_1.01,1.21910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,2,power_law_1.2,0.35012478828430177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,2,power_law_1.2,1.4416128158569337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,2,power_law_1.01,0.575878381729126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,2,power_law_1.01,4.1445568084716795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,2,power_law_1.2,0.41341438293457033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,2,power_law_1.01,1.2404735565185547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,2,power_law_1.01,8.811917114257813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,2,power_law_1.01,0.62674560546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,2,power_law_1.2,1.5187711715698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,2,power_law_1.2,0.7091392040252685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,2,1,balanced,0.07857066889603932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,2,power_law_1.01,1.232819175720215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,2,1,balanced,0.1471733351548513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,2,1,balanced,0.24109333753585815
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,2,power_law_1.01,0.7448895931243896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,2,1,balanced,0.40005866686503094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,2,power_law_1.2,0.7353216171264648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,2,power_law_1.2,0.08916479945182801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,2,1,balanced,0.7325173219045004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,2,power_law_1.2,1.8323583602905273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,2,power_law_1.01,1.3198464393615723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,2,power_law_1.01,0.8463808059692383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,2,power_law_1.2,0.6965824127197265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,2,power_law_1.2,0.13959039449691774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,2,1,balanced,1.4187893867492676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,2,power_law_1.2,0.19212160110473633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,2,power_law_1.01,5.387398529052734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,2,power_law_1.2,0.7208000183105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,2,power_law_1.2,0.05167359709739685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,2,power_law_1.01,1.0766271591186523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,2,power_law_1.2,1.9812543869018555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,2,power_law_1.2,0.2693183898925781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,2,power_law_1.2,0.07200639843940734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,2,power_law_1.2,0.717529582977295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,2,power_law_1.2,0.37521278858184814
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,2,power_law_1.2,0.09222400188446045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,2,power_law_1.2,0.7468480110168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,2,power_law_1.2,0.14339840412139893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,2,power_law_1.2,0.7570879936218262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,2,power_law_1.2,0.18460160493850708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,2,power_law_1.2,0.8056447982788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,2,power_law_1.2,0.20484480857849122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,2,power_law_1.01,1.3961088180541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,2,1,balanced,1.4640159606933594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,2,power_law_1.2,0.8081855773925781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,2,power_law_1.2,0.31398398876190187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,2,1,balanced,0.05622933308283488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,2,1,balanced,0.07724266747633617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,2,1,balanced,1.45742400487264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,2,1,balanced,0.12692266702651978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,2,power_law_1.2,0.8259776115417481
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,2,power_law_1.2,0.3283456087112427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,2,power_law_1.01,1.55282564163208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,2,power_law_1.01,1.3248064041137695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,2,1,balanced,1.5892799695332844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,2,power_law_1.2,0.355295991897583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,2,power_law_1.2,0.8536895751953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,2,power_law_1.2,2.1648000717163085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,2,1,balanced,1.5897919336954753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,2,power_law_1.01,1.7429695129394531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,2,power_law_1.2,0.34690558910369873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,2,power_law_1.2,1.0430399894714355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,2,power_law_1.01,1.7772287368774413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,2,1,balanced,1.5723519325256348
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,2,power_law_1.2,0.37392001152038573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,2,power_law_1.2,2.3047040939331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,2,power_law_1.2,1.1339455604553224
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,2,power_law_1.2,0.382751989364624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,2,1,balanced,1.587007999420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,2,power_law_1.01,2.1001728057861326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,2,power_law_1.2,0.40166401863098145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,2,1,balanced,0.2237493395805359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,2,1,balanced,1.598533312479655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,2,1,balanced,0.4116906722386678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,2,power_law_1.2,1.276966381072998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,2,power_law_1.2,0.45122561454772947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,2,1,balanced,0.7761226495107015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,2,power_law_1.2,2.810905647277832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,2,1,balanced,1.6060959498087566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,2,power_law_1.01,2.462784004211426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,2,power_law_1.2,0.4236032009124756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,2,1,balanced,0.9198079903920492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,2,power_law_1.2,1.4436927795410157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,2,1,balanced,1.607082684834798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,2,1,balanced,0.921504020690918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,2,power_law_1.2,0.4278656005859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,2,1,balanced,0.8521866798400879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,2,power_law_1.2,0.4856704235076904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,2,1,balanced,0.854085365931193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,2,power_law_1.2,1.6951936721801757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,2,power_law_1.2,0.44759039878845214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,2,power_law_1.2,3.2653057098388674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,2,power_law_1.2,0.4946944236755371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,2,1,balanced,0.8054506778717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,2,power_law_1.01,3.1983680725097656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,2,1,balanced,0.8154719670613607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,2,power_law_1.2,0.656819200515747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,2,power_law_1.2,0.554252815246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,2,1,balanced,0.8166133562723795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,2,power_law_1.01,3.1340160369873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,2,power_law_1.2,2.0014720916748048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,2,1,balanced,0.8243839740753174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,2,power_law_1.2,0.5981696128845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,2,power_law_1.2,0.6026112079620362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,2,1,balanced,0.8338133494059244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,2,1,balanced,0.8420213063557943
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,2,power_law_1.2,0.6674431800842285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,2,power_law_1.2,4.289888000488281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,2,power_law_1.01,3.8526016235351563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,2,power_law_1.2,2.562156867980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,2,1,balanced,1.6110666592915852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,2,power_law_1.2,0.781273603439331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,2,1,balanced,1.5930399894714355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,2,power_law_1.2,0.9405695915222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,2,1,balanced,1.8413813908894856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,2,power_law_1.2,3.2365440368652343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,2,1,balanced,0.06778133412202199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,2,1,balanced,0.08787733316421509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,2,1,balanced,1.866693337758382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,2,1,balanced,0.13135466972986856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,2,power_law_1.2,1.1296640396118165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,2,1,balanced,0.2301386594772339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,2,power_law_1.2,5.2701568603515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,2,power_law_1.01,5.3499198913574215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,2,1,balanced,0.4314773480097453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,2,1,balanced,0.8570559819539388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,2,1,balanced,1.9596053759257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,2,1,balanced,1.051743984222412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,2,power_law_1.2,1.419929599761963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,2,1,balanced,2.0042346318562827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,2,1,balanced,1.0720799763997395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,2,power_law_1.2,4.413894271850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,2,1,balanced,1.1191360155741374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,2,1,balanced,2.206127961476644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,2,1,balanced,1.1690080165863037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,2,1,balanced,1.273525317509969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,2,1,balanced,2.3039093017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,2,1,balanced,0.8023839791615804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,2,1,balanced,1.3876694043477376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,2,power_law_1.01,6.721612548828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,2,power_law_1.2,7.448518371582031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,2,1,balanced,0.8049919605255127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,2,1,balanced,0.8056853612263998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,2,power_law_1.2,5.532358551025391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,2,1,balanced,2.1446827252705893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,2,1,balanced,3.6299146016438804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,2,1,balanced,1.3433760007222493
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,2,power_law_1.2,1.8460607528686523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,2,1,balanced,2.3708693186442056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,2,1,balanced,1.349562644958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,2,1,balanced,1.3585386276245117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,2,1,balanced,4.011573473612468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,2,1,balanced,1.362613360087077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,2,1,balanced,3.3972959518432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,2,1,balanced,1.3679572741190593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,2,power_law_1.2,3.432582473754883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,2,1,balanced,1.3807306289672852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,2,1,power_law_1.01,0.06709120273590088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,2,1,balanced,5.69822374979655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,2,power_law_1.2,9.05994873046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,2,power_law_1.01,9.564141082763673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,2,1,balanced,1.393082618713379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,2,1,power_law_1.01,0.12940160036087037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,2,1,balanced,1.4054826100667317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,2,1,power_law_1.01,0.2108544111251831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,2,1,power_law_1.01,0.05545600056648255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,2,1,balanced,1.4262399673461914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,2,1,power_law_1.01,0.07834879755973816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,2,1,power_law_1.01,0.13007999658584596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,2,1,balanced,1.463311990102132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,2,power_law_1.2,1.0369152069091796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,2,1,power_law_1.01,0.22113919258117676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,2,1,balanced,1.496613343556722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,2,1,balanced,7.36192003885905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,2,power_law_1.2,1.0650943756103515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,2,1,balanced,4.4063412348429365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,2,1,balanced,1.569658597310384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,2,power_law_1.2,1.1857088088989258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,2,1,balanced,1.6505653063456218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,2,1,balanced,0.032373333970705666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,2,1,balanced,0.05268799761931101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,2,1,power_law_1.01,0.3699392080307007
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,2,1,balanced,0.0765066643555959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,2,power_law_1.2,1.1703680038452149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,2,1,power_law_1.01,0.5495999813079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,2,1,balanced,2.7944533030192056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,2,power_law_1.2,1.2100671768188476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,2,1,power_law_1.01,0.7492608070373535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,2,1,power_law_1.01,0.3231359958648682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,2,power_law_1.01,12.80865936279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,2,power_law_1.2,1.2673664093017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,2,1,balanced,2.937525431315104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,2,1,power_law_1.01,0.4351679801940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,2,1,power_law_1.01,1.0748543739318848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,2,1,power_law_1.2,0.06668800115585327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,2,1,power_law_1.01,0.6675712108612061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,2,power_law_1.2,1.31911039352417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,2,1,power_law_1.2,0.1285375952720642
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,2,1,balanced,0.12363732854525249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,2,1,power_law_1.01,1.126796817779541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,2,1,balanced,0.21619200706481934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,2,1,power_law_1.01,0.7021247863769531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,2,1,power_law_1.2,0.21130878925323487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,2,1,balanced,0.40223467350006104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,2,power_law_1.2,1.3736063957214355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,2,1,balanced,4.266895929972331
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,2,1,balanced,0.40454399585723877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,2,1,power_law_1.01,1.2234623908996582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,2,1,balanced,0.4121280113855998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,2,1,power_law_1.2,0.36584959030151365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,2,1,balanced,0.4148053328196208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,2,1,balanced,0.4161546627680461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,2,power_law_1.2,1.482643222808838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,2,1,power_law_1.2,0.5061247825622559
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,2,1,balanced,0.420634667078654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,2,1,power_law_1.01,1.2730048179626465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,2,1,balanced,0.4264320135116577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,2,1,balanced,0.4325706561406453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,2,1,power_law_1.2,0.6730751991271973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,2,1,balanced,0.44489598274230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,2,power_law_1.2,1.652569580078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,2,1,balanced,0.4746026595433553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,2,1,balanced,5.566229502360026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,2,1,balanced,0.48104000091552734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,2,1,power_law_1.2,1.0618111610412597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,2,1,balanced,0.5068853298823038
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,2,1,balanced,0.5111413399378458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,2,power_law_1.2,1.8321727752685546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,2,1,balanced,0.5384693145751953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,2,1,power_law_1.2,1.0988224029541016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,2,1,balanced,0.5888693332672119
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,2,1,balanced,0.6326239903767904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,2,1,power_law_1.01,0.03256320059299469
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,2,1,balanced,0.8037066459655762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,2,1,power_law_1.2,1.2342464447021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,2,power_law_1.2,2.2317119598388673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,2,1,power_law_1.01,0.05331839919090271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,2,1,balanced,0.8139839967091879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,2,1,power_law_1.01,0.07665280103683472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,2,1,balanced,8.14901860555013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,2,1,power_law_1.2,1.2531007766723632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,2,1,balanced,1.6079413096110027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,2,1,power_law_1.01,0.12354559898376465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,2,1,balanced,1.149733304977417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,2,power_law_1.2,2.595648002624512
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,2,1,power_law_1.01,0.173471999168396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,2,1,power_law_1.01,0.6763711929321289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,2,1,balanced,1.6291252772013347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,2,1,power_law_1.01,0.23322880268096924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,2,1,power_law_1.01,0.6976384162902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,2,1,power_law_1.01,0.3146687984466553
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,2,1,balanced,2.104309399922689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,2,1,power_law_1.01,0.6773759841918945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,2,1,power_law_1.01,0.06422399878501892
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,2,1,power_law_1.01,0.3250047922134399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,2,power_law_1.2,3.4563392639160155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,2,1,power_law_1.01,1.2826815605163575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,2,1,power_law_1.01,0.694643211364746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,2,1,power_law_1.01,0.3310719966888428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,2,1,power_law_1.01,0.708083200454712
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,2,1,power_law_1.01,0.3388351917266846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,2,1,power_law_1.01,1.281868839263916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,2,1,power_law_1.2,1.2482815742492677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,2,1,balanced,10.740922292073568
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,2,1,power_law_1.01,0.3531584024429321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,2,1,power_law_1.01,0.7229887962341308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,2,1,power_law_1.01,0.3647871971130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,2,1,power_law_1.2,1.287936019897461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,2,power_law_1.2,4.1586559295654295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,2,1,power_law_1.01,0.37287681102752684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,2,1,power_law_1.2,1.3298239707946777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,2,1,power_law_1.01,0.40662398338317873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,2,1,power_law_1.2,0.052832001447677614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,2,1,power_law_1.01,0.44021759033203123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,2,1,power_law_1.2,0.0754751980304718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,2,1,power_law_1.2,1.3627776145935058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,2,1,power_law_1.2,0.12673280239105225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,2,1,power_law_1.01,0.4703104019165039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,2,1,power_law_1.2,0.2099519968032837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,2,1,power_law_1.2,1.3994815826416016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,2,1,power_law_1.01,0.5441792011260986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,2,1,power_law_1.01,1.2994303703308105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,2,power_law_1.2,5.600543975830078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,2,1,power_law_1.01,0.736959981918335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,2,1,power_law_1.2,0.06503040194511414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,2,1,power_law_1.2,0.2912192106246948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,2,1,power_law_1.01,0.5200704097747803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,2,1,power_law_1.2,0.0870464026927948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,2,1,power_law_1.2,1.4376511573791504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,2,1,power_law_1.01,0.7609600067138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,2,1,power_law_1.2,0.38099839687347414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,2,1,power_law_1.01,1.366163158416748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,2,1,power_law_1.01,0.6074751853942871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,2,1,power_law_1.2,0.1306496024131775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,2,1,power_law_1.01,0.7831871986389161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,2,1,power_law_1.2,0.6471424102783203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,2,1,power_law_1.2,0.2195136070251465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,2,1,power_law_1.01,0.6138879776000976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,2,1,power_law_1.2,1.4958592414855958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,2,1,power_law_1.01,1.3726143836975098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,2,1,power_law_1.01,0.9848896026611328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,2,1,power_law_1.2,0.7024511814117431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,2,1,power_law_1.01,0.7302527904510498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,2,1,power_law_1.01,1.4014592170715332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,2,1,power_law_1.2,1.8641279220581055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,2,1,power_law_1.01,1.0646719932556152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,2,1,power_law_1.2,0.6756288051605225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,2,1,power_law_1.01,0.7778495788574219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,2,1,power_law_1.01,0.08648319840431214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,2,1,power_law_1.2,0.6874752044677734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,2,power_law_1.2,7.130528259277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,2,1,power_law_1.01,1.1580991744995117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,2,1,power_law_1.01,1.4452799797058105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,2,1,power_law_1.01,0.12992639541625978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,2,1,power_law_1.2,1.8923391342163085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,2,1,power_law_1.2,0.6684095859527588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,2,1,power_law_1.01,0.22844159603118896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,2,1,power_law_1.01,1.2588095664978027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,2,1,power_law_1.2,0.7048960208892823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,2,1,power_law_1.01,0.3334208011627197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,2,1,power_law_1.01,1.768172836303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,2,1,power_law_1.2,0.3083264112472534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,2,1,power_law_1.2,2.0674943923950195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,2,1,power_law_1.2,0.6841919898986817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,2,1,power_law_1.01,1.4965951919555665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,2,1,power_law_1.2,0.4196415901184082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,2,1,power_law_1.2,0.7331456184387207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,2,1,power_law_1.01,1.8370752334594727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,2,1,power_law_1.2,0.6006527900695801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,2,1,power_law_1.01,1.7528255462646485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,2,1,power_law_1.2,0.753062391281128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,2,1,power_law_1.01,0.9316415786743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,2,1,power_law_1.2,2.2628032684326174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,2,1,power_law_1.2,0.6165760040283204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,2,1,power_law_1.2,0.7648640155792237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,2,1,power_law_1.01,2.0074304580688476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,2,1,power_law_1.01,1.0883968353271485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,2,1,power_law_1.2,0.8183423995971679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,2,1,power_law_1.2,1.0759039878845216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,2,1,power_law_1.2,2.7067583084106444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,2,power_law_1.2,9.881011199951171
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,2,1,power_law_1.01,1.331123161315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,2,1,power_law_1.2,1.0165568351745606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,2,1,power_law_1.2,1.0857215881347657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,2,1,power_law_1.2,1.084864044189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,2,1,power_law_1.2,1.1294400215148925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,2,1,power_law_1.01,1.8154176712036132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,2,1,power_law_1.2,3.0338815689086913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,2,1,power_law_1.2,1.2177472114562988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,2,1,power_law_1.2,1.137548828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,2,1,power_law_1.01,2.2536127090454103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,2,1,power_law_1.2,1.1777407646179199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,2,1,power_law_1.2,1.3315391540527344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,2,1,power_law_1.01,2.7726144790649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,2,1,power_law_1.01,2.177996826171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,2,1,power_law_1.2,3.8060222625732423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,2,1,power_law_1.2,1.2204671859741212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,2,1,power_law_1.2,1.5674176216125488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,2,1,power_law_1.01,2.6931711196899415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,2,1,power_law_1.01,2.534111976623535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,2,1,power_law_1.2,1.2975296020507812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,2,1,power_law_1.2,1.8081151962280273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,2,1,power_law_1.01,0.44159998893737795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,2,1,power_law_1.2,1.3421055793762207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,2,1,power_law_1.01,0.6195136070251465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,2,1,power_law_1.01,3.687052917480469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,2,1,power_law_1.2,4.6004993438720705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,2,power_law_1.2,12.834072875976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,2,1,power_law_1.01,2.987411117553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,2,1,power_law_1.2,2.2899520874023436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,2,1,power_law_1.2,1.432089614868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,2,1,power_law_1.01,0.6540224075317382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,2,1,power_law_1.2,1.5859583854675292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,2,1,power_law_1.2,2.810041618347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,2,1,power_law_1.01,3.773388671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,2,1,power_law_1.01,4.7665153503417965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,2,1,power_law_1.2,1.7666879653930665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,2,1,power_law_1.2,6.519980621337891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,2,1,power_law_1.2,3.8033729553222657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,2,1,power_law_1.2,2.093395233154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,2,1,power_law_1.01,4.460083389282227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,2,1,power_law_1.2,2.4164031982421874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,2,balanced,0.02903999884923299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,2,balanced,0.03530666728814443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,2,1,power_law_1.2,0.0324288010597229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,2,balanced,0.0533493310213089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,2,balanced,0.07655466596285503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,2,balanced,0.1246613363424937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,2,balanced,0.12687466541926065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,2,1,power_law_1.2,4.731731033325195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,2,1,power_law_1.2,0.05300480127334595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,2,balanced,0.13590400417645773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,2,balanced,0.13704533378283182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,2,balanced,0.14403733611106873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,2,balanced,0.14197333653767905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,2,1,power_law_1.2,0.07715839743614197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,2,balanced,0.135861337184906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,2,balanced,0.13583466410636902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,2,1,power_law_1.2,7.729452514648438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,2,1,power_law_1.2,0.11445120573043824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,2,1,power_law_1.01,6.144134521484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,2,1,power_law_1.2,0.15880320072174073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,2,1,power_law_1.2,0.22001919746398926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,2,balanced,0.027274665733178455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,2,balanced,0.03173866619666418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,2,balanced,0.039434666434923805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,2,balanced,0.05555733541647593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,2,1,power_law_1.2,0.2999039888381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,2,balanced,0.08719999591509502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,2,balanced,0.08893866340319316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,2,balanced,0.09828799962997437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,2,balanced,0.09778133034706116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,2,1,power_law_1.2,0.33029119968414306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,2,balanced,0.10057600339253743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,2,balanced,0.1009386678536733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,2,balanced,0.09713066617647807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,2,1,power_law_1.2,3.0202495574951174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,2,1,power_law_1.2,0.33968639373779297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,2,balanced,0.025455998877684276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,2,balanced,0.02757866680622101
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,2,1,power_law_1.2,0.3400768041610718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,2,balanced,0.03065599997838338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,2,balanced,0.03743999948104223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,2,balanced,0.05484800040721893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,2,balanced,0.05644266804059347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,2,1,power_law_1.2,0.3567935943603516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,2,balanced,0.06019733349482218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,2,balanced,0.06237866481145223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,2,balanced,0.06239999830722809
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,2,balanced,0.06337066491444905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,2,1,power_law_1.01,7.881318664550781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,2,balanced,0.06503466765085857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,2,balanced,0.062224000692367554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,2,balanced,0.06394133468468984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,2,1,power_law_1.2,3.6744449615478514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,2,balanced,0.06614399949709575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,2,balanced,0.06540266672770183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,2,balanced,0.06611733138561249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,2,balanced,0.07014399766921997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,2,balanced,0.0766133318344752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,2,balanced,0.07943999767303467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,2,balanced,0.09852266311645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,2,balanced,0.09819733103116353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,2,balanced,0.136543999115626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,2,balanced,0.10311466455459595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,2,balanced,0.1397706667582194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,2,balanced,0.15307199954986572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,2,balanced,0.14268799622853598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,2,balanced,0.1481119990348816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,2,balanced,0.15498666961987814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,2,balanced,0.18983999888102213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,2,balanced,0.20317333936691284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,2,balanced,0.22706133127212524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,2,balanced,0.2529066602389018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,2,balanced,0.03573333223660787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,2,balanced,0.037690666814645134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,2,1,power_law_1.2,0.3723328113555908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,2,1,power_law_1.2,4.873158264160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,2,balanced,0.045370668172836304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,2,balanced,0.06403733293215434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,2,balanced,0.09154133001963298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,2,balanced,0.09572266538937886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,2,1,power_law_1.2,0.3810816049575806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,2,balanced,0.0978666643301646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,2,1,power_law_1.01,1.0807104110717773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,2,balanced,0.09750399986902873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,2,balanced,0.13773333032925925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,2,balanced,0.1381760040918986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,2,1,power_law_1.2,0.4264768123626709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,2,balanced,0.14041067163149515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,2,balanced,0.1426293353239695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,2,balanced,0.0993226667245229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,2,1,power_law_1.2,0.44488959312438964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,2,balanced,0.16132799784342447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,2,balanced,0.22405866781870523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,2,balanced,0.28441067536671955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,2,balanced,0.40059733390808105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,2,balanced,0.5241599877675375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,2,balanced,0.3689546585083008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,2,power_law_1.01,0.03131519854068756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,2,1,power_law_1.2,6.069030380249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,2,balanced,0.4206399917602539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,2,power_law_1.01,0.030432000756263733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,2,balanced,0.5853066841761271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,2,power_law_1.01,0.04500479996204376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,2,power_law_1.01,0.036294400691986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,2,balanced,0.7497066656748453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,2,power_law_1.01,0.04444159865379334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,2,power_law_1.01,0.04962559938430786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,2,1,power_law_1.01,1.097856044769287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,2,balanced,0.10187733173370361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,2,balanced,0.10731732845306396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,2,power_law_1.01,0.05986559987068176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,2,balanced,0.11226133505503337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,2,balanced,0.11845866839090984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,2,power_law_1.01,0.08421760201454162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,2,1,power_law_1.2,0.48903679847717285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,2,balanced,0.14683199922243753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,2,1,power_law_1.01,1.1571776390075683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,2,power_law_1.01,0.09121919870376587
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,2,1,power_law_1.2,0.5716608047485352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,2,power_law_1.01,0.09182720184326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,2,power_law_1.01,0.09496960043907166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,2,1,power_law_1.2,0.5398208141326905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,2,power_law_1.01,0.09447680115699768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,2,power_law_1.01,0.09475200176239014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,2,1,power_law_1.2,0.6335999965667725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,2,power_law_1.01,0.055718398094177245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,2,power_law_1.01,0.09591680169105529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,2,power_law_1.01,0.07304959893226623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,2,power_law_1.01,0.09761919975280761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,2,balanced,0.14257066448529562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,2,1,power_law_1.2,0.6454976081848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,2,1,power_law_1.2,8.422361755371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,2,power_law_1.01,0.08084480166435241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,2,power_law_1.01,0.10196479558944702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,2,power_law_1.01,0.11383039951324463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,2,power_law_1.01,0.10721280574798583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,2,1,power_law_1.2,0.7681087970733642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,2,power_law_1.01,0.12416640520095826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,2,balanced,0.16169066230456033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,2,power_law_1.01,0.11390719413757325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,2,balanced,0.19131733973821005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,2,power_law_1.01,0.1247488021850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,2,balanced,0.21875733137130737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,2,power_law_1.01,0.12584960460662842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,2,balanced,0.3142933249473572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,2,power_law_1.01,0.13781119585037233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,2,1,power_law_1.2,0.8038144111633301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,2,power_law_1.01,0.15738879442214965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,2,balanced,0.3735733429590861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,2,1,power_law_1.01,1.1690752029418945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,2,power_law_1.01,0.14042240381240845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,2,balanced,0.522704005241394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,2,power_law_1.01,0.1756608009338379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,2,power_law_1.01,0.13051520586013793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,2,1,power_law_1.2,0.9503935813903809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,2,balanced,0.6735946337381998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,2,power_law_1.01,0.21488640308380128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,2,power_law_1.01,0.13313920497894288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,2,1,power_law_1.01,1.157209587097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,2,power_law_1.01,0.25173120498657225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,2,power_law_1.01,0.13358080387115479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,2,balanced,0.14827733238538107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,2,1,power_law_1.2,1.1165504455566406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,2,power_law_1.01,0.13664640188217164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,2,power_law_1.01,0.3279936075210571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,2,1,power_law_1.01,1.2392191886901855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,2,power_law_1.01,0.40853118896484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,2,1,power_law_1.2,1.364793586730957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,2,power_law_1.01,0.5681215763092041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,2,1,power_law_1.01,1.2590911865234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,2,power_law_1.01,0.7301951885223389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,2,1,power_law_1.01,1.278335952758789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,2,1,power_law_1.2,10.730271911621093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,2,1,power_law_1.2,1.844268798828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,2,balanced,0.154831995566686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,2,1,power_law_1.01,1.3794495582580566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,2,balanced,0.15902400016784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,2,balanced,0.16876266400019327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,2,power_law_1.01,0.13952640295028687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,2,balanced,0.18549867471059164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,2,balanced,0.20779200394948324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,2,power_law_1.01,0.1505280017852783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,2,balanced,0.3170773386955261
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,2,1,power_law_1.2,2.928985595703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,2,power_law_1.01,0.16586240530014038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,2,1,power_law_1.01,1.5395520210266114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,2,power_law_1.01,0.2035968065261841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,2,power_law_1.01,0.222707200050354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,2,power_law_1.01,0.26301438808441163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,2,1,power_law_1.01,1.6756671905517577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,2,power_law_1.01,0.3085504055023193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,2,power_law_1.01,0.38385920524597167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,2,1,power_law_1.01,1.9859968185424806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,2,power_law_1.01,0.46913919448852537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,2,balanced,0.3516800006230672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,2,power_law_1.01,0.6672895908355713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,2,balanced,0.49564798672993976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,2,power_law_1.01,0.025600001215934753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,2,balanced,0.6401120026906332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,2,power_law_1.01,0.8317567825317382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,2,1,power_law_1.01,2.335251235961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,2,power_law_1.01,0.02800000011920929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,2,balanced,0.9218453566233317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,2,power_law_1.01,0.0315775990486145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,2,balanced,1.2159360249837239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,2,power_law_1.01,0.03553920090198517
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,2,power_law_1.01,0.04022400081157684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,2,balanced,1.7628320058186848
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,2,power_law_1.01,0.05435519814491272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,2,balanced,1.0819199879964192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,2,1,power_law_1.01,2.9648576736450196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,2,power_law_1.01,0.05735679864883423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,2,balanced,1.431338628133138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,2,power_law_1.01,0.058841598033905027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,2,balanced,2.3457013765970864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,2,power_law_1.01,0.05978879928588867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,2,power_law_1.2,0.03165439963340759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,2,power_law_1.2,0.02942720055580139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,2,power_law_1.01,0.0605567991733551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,2,power_law_1.2,0.03590399920940399
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,2,power_law_1.01,0.06410880088806152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,2,power_law_1.2,0.041945600509643556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,2,power_law_1.01,0.06111999750137329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,2,1,power_law_1.01,3.6034751892089845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,2,power_law_1.2,0.04631040096282959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,2,power_law_1.01,0.06279680132865906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,2,power_law_1.2,0.05605120062828064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,2,power_law_1.01,0.06555520296096802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,2,power_law_1.2,0.0843775987625122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,2,power_law_1.01,0.06596480011940002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,2,power_law_1.2,0.08720639944076539
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,2,power_law_1.01,0.06954240202903747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,2,power_law_1.2,0.09182080030441284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,2,power_law_1.01,0.0712768018245697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,2,power_law_1.2,0.0927295982837677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,2,power_law_1.2,0.09257599711418152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,2,power_law_1.2,0.03932160139083862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,2,power_law_1.2,0.09488000273704529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,2,power_law_1.2,0.056966400146484374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,2,1,power_law_1.01,4.864601516723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,2,power_law_1.2,0.09615359902381897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,2,power_law_1.2,0.06714879870414733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,2,power_law_1.2,0.0986240029335022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,2,power_law_1.2,0.07150080204010009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,2,power_law_1.2,0.1015936017036438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,2,power_law_1.2,0.1102720022201538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,2,power_law_1.2,0.1102720022201538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,2,power_law_1.2,0.12394239902496337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,2,power_law_1.2,0.11677440404891967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,2,power_law_1.2,0.12620160579681397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,2,power_law_1.2,0.12776960134506227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,2,power_law_1.2,0.13834240436553955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,2,power_law_1.2,0.15541759729385377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,2,power_law_1.2,0.13870079517364503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,2,power_law_1.01,0.07826560139656066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,2,power_law_1.2,0.13175040483474731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,2,power_law_1.2,0.17852799892425536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,2,power_law_1.01,0.08790400028228759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,2,power_law_1.2,0.13152639865875243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,2,power_law_1.2,0.21860480308532715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,2,1,power_law_1.01,6.097401428222656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,2,power_law_1.01,0.10993280410766601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,2,power_law_1.2,0.13279999494552613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,2,power_law_1.2,0.2583872079849243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,2,power_law_1.2,0.13731199502944946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,2,power_law_1.2,0.3391040086746216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,2,power_law_1.01,0.036032000184059144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,2,power_law_1.2,0.4356287956237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,2,power_law_1.01,0.04444800019264221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,2,power_law_1.01,0.05121279954910278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,2,power_law_1.01,1.0337984085083007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,2,power_law_1.01,0.05816320180892944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,2,balanced,0.9672160148620605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,2,power_law_1.01,0.06551679968833923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,2,power_law_1.01,0.08748800158500672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,2,power_law_1.01,1.3405440330505372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,2,balanced,1.2822613716125488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,2,power_law_1.01,0.09137279987335205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,2,power_law_1.01,0.12406400442123414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,2,power_law_1.01,0.09082239866256714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,2,power_law_1.2,0.036268800497055054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,2,power_law_1.2,0.14181120395660402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,2,power_law_1.01,0.13296639919281006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,2,power_law_1.01,0.1569408059120178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,2,power_law_1.2,0.15727360248565675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,2,power_law_1.01,0.1347584009170532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,2,1,power_law_1.01,8.530387115478515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,2,power_law_1.01,0.18616960048675538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,2,power_law_1.2,0.16836479902267457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,2,power_law_1.01,0.13703680038452148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,2,power_law_1.2,0.2093183994293213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,2,power_law_1.01,0.24865920543670655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,2,power_law_1.01,0.1396415948867798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,2,power_law_1.2,0.2284480094909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,2,power_law_1.01,0.14124159812927245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,2,power_law_1.2,0.2727679967880249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,2,power_law_1.01,0.1498047947883606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,2,power_law_1.2,0.32063999176025393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,2,power_law_1.01,0.15828479528427125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,2,power_law_1.01,0.16792960166931153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,2,power_law_1.2,0.02353920042514801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,2,power_law_1.01,0.18961280584335327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,2,power_law_1.01,1.1906304359436035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,2,power_law_1.2,0.027039998769760133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,2,power_law_1.2,0.04270080029964447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,2,power_law_1.01,0.22750720977783204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,2,power_law_1.2,0.02963840067386627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,2,power_law_1.2,0.05004799962043762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,2,power_law_1.01,0.26296319961547854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,2,power_law_1.2,0.03407999873161316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,2,power_law_1.2,0.05613440275192261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,2,power_law_1.01,0.3321727991104126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,2,power_law_1.2,0.040140798687934874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,2,power_law_1.01,0.31680641174316404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,2,power_law_1.2,0.5877312183380127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,2,power_law_1.2,0.05253120064735413
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,2,power_law_1.01,0.4405375957489014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,2,power_law_1.2,0.056364798545837404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,2,power_law_1.2,0.4011392116546631
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,2,power_law_1.2,0.05788159966468811
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,2,power_law_1.01,0.5889791965484619
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,2,power_law_1.2,0.05948160290718078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,2,1,balanced,0.023029332359631855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,2,1,balanced,0.031093334158261616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,2,power_law_1.2,0.4868480205535889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,2,1,balanced,0.0451200008392334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,2,power_law_1.2,0.06111360192298889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,2,1,balanced,0.07891733447710673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,2,1,balanced,0.13185066978136697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,2,1,balanced,0.13529066244761148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,2,power_law_1.2,0.06409599781036376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,2,1,balanced,0.141184002161026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,2,1,balanced,0.14056533575057983
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,2,power_law_1.2,0.06104320287704468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,2,1,balanced,0.1418186624844869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,2,1,balanced,0.14226667086283365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,2,power_law_1.01,1.5227968215942382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,2,1,balanced,0.1368160049120585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,2,power_law_1.2,0.06367999911308289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,2,1,balanced,0.13481066624323526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,2,power_law_1.2,0.06358399987220764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,2,1,balanced,0.13526933391888937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,2,power_law_1.2,0.06606079936027527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,2,1,balanced,0.13768532872200012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,2,power_law_1.01,0.41152000427246094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,2,power_law_1.2,0.06484479904174804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,2,power_law_1.2,0.06783999800682068
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,2,power_law_1.2,0.07303040027618408
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,2,power_law_1.2,0.08373759984970093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,2,1,power_law_1.01,10.885664367675782
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,2,power_law_1.2,0.08926079869270324
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,2,power_law_1.2,0.11160320043563843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,2,1,balanced,0.02942933390537898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,2,power_law_1.2,0.1308351993560791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,2,1,balanced,0.03123733401298523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,2,1,balanced,0.02256533255179723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,2,1,balanced,0.039706667264302574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,2,1,balanced,0.02532800038655599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,2,1,balanced,0.06039466460545858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,2,1,balanced,0.03136533250411352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,2,1,balanced,0.09610666831334432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,2,power_law_1.2,0.16546560525894166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,2,1,balanced,0.13763200243314108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,2,1,balanced,0.14062399665514627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,2,1,balanced,0.1458453337351481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,2,1,balanced,0.18184532721837363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,2,power_law_1.01,0.5569471836090087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,2,1,balanced,0.18878400325775146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,2,1,balanced,0.20596800247828165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,2,1,balanced,0.22450133164723715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,2,1,balanced,0.33477866649627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,2,power_law_1.01,0.7103871822357177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,2,1,balanced,0.3707840045293172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,2,1,balanced,0.512661337852478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,2,power_law_1.01,0.9928768157958985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,2,1,balanced,0.6635733445485433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,2,power_law_1.2,0.6757696151733399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,2,1,balanced,0.9586079915364584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,2,1,balanced,0.056032001972198486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,2,power_law_1.01,1.277843189239502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,2,power_law_1.2,0.08903040289878845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,2,power_law_1.2,0.7319424152374268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,2,1,balanced,1.257695992787679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,2,power_law_1.2,0.089683198928833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,2,power_law_1.2,0.09190400242805481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,2,power_law_1.2,0.1336127996444702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,2,power_law_1.01,1.9087488174438476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,2,power_law_1.2,0.1351680040359497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,2,power_law_1.2,0.13751039505004883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,2,power_law_1.2,0.14018559455871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,2,power_law_1.2,0.1989184021949768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,2,power_law_1.2,0.14338560104370118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,2,power_law_1.01,2.442950439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,2,1,balanced,0.09825600186983745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,2,power_law_1.2,0.15329279899597167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,2,1,power_law_1.01,0.021740800142288207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,2,1,balanced,0.09630933403968811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,2,1,balanced,0.09982400139172871
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,2,1,balanced,0.025813333690166473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,2,power_law_1.2,0.1624768018722534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,2,1,power_law_1.01,0.031385600566864014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,2,1,balanced,0.027717334528764088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,2,1,balanced,0.03027733415365219
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,2,1,balanced,0.03505599995454153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,2,power_law_1.2,0.1806015968322754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,2,1,balanced,0.055813332398732506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,2,1,balanced,0.05783466498057047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,2,1,balanced,0.05955199897289276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,2,1,balanced,0.061237335205078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,2,1,balanced,0.06275199850400288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,2,1,balanced,0.06368533273537953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,2,1,balanced,0.06500266492366791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,2,1,balanced,0.06830400228500366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,2,1,balanced,0.06947733461856842
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,2,1,balanced,0.07425066828727722
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,2,1,balanced,0.07241599758466084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,2,power_law_1.2,0.2750272035598755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,2,1,balanced,0.07238933444023132
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,2,1,balanced,0.07845333218574524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,2,1,balanced,0.08354133367538452
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,2,1,balanced,0.09085866808891296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,2,1,balanced,0.10840533177057902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,2,1,balanced,0.08747733632723491
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,2,1,balanced,0.112527996301651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,2,1,balanced,0.08873599767684937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,2,1,balanced,0.09805333614349365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,2,1,balanced,0.19785600900650024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,2,1,balanced,0.09912533561388652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,2,1,balanced,0.1018293301264445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,2,1,balanced,0.18530666828155518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,2,1,balanced,0.09775466720263164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,2,1,balanced,0.09880533814430237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,2,1,balanced,0.1034879982471466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,2,1,balanced,0.2563093304634094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,2,1,balanced,0.10123733679453532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,2,1,balanced,0.09909866253534953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,2,1,balanced,0.10342400272687276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,2,1,balanced,0.09790399670600891
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,2,1,balanced,0.3210666577021281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,2,1,balanced,0.10381866494814555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,2,1,balanced,0.09941333532333374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,2,power_law_1.2,0.19309439659118652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,2,1,balanced,0.10129066308339436
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,2,1,balanced,0.4497973521550496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,2,1,balanced,0.10250666737556458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,2,1,balanced,0.10685867071151733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,2,power_law_1.2,0.23752319812774658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,2,1,balanced,0.10960533221562703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,2,1,balanced,0.14038399855295816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,2,1,balanced,0.15191466609636942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,2,power_law_1.2,0.26067841053009033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,2,1,balanced,0.16955200831095377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,2,1,balanced,0.1886026660601298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,2,power_law_1.2,0.34607999324798583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,2,power_law_1.2,0.8648768424987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,2,1,power_law_1.01,0.04562560021877289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,2,power_law_1.2,0.4137728214263916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,2,1,power_law_1.01,0.06919040083885193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,2,1,balanced,0.10717866818110149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,2,1,balanced,0.1351093351840973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,2,1,balanced,0.13770133256912231
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,2,1,balanced,0.5859573284784952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,2,1,balanced,0.15897599856058756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,2,1,balanced,0.1711519956588745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,2,1,balanced,0.18278400103251138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,2,1,power_law_1.01,0.0220223993062973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,2,1,balanced,0.28329066435496014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,2,1,balanced,0.3083786765734355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,2,1,balanced,0.2797173261642456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,2,1,balanced,0.3158559997876485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,2,1,balanced,0.4382133483886719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,2,1,balanced,0.5680160125096639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,2,power_law_1.2,0.5787519931793212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,2,1,balanced,0.8214080333709717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,2,1,balanced,1.0834986368815105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,2,1,power_law_1.01,0.07897599935531616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,2,1,power_law_1.01,0.11352959871292115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,2,1,power_law_1.01,0.12878719568252564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,2,1,power_law_1.01,0.026393601298332216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,2,1,power_law_1.01,0.1290176033973694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,2,1,power_law_1.01,0.03340800106525421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,2,power_law_1.2,0.7230400085449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,2,1,power_law_1.01,0.13327360153198242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,2,1,power_law_1.01,0.04830079972743988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,2,1,power_law_1.01,0.13386240005493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,2,1,power_law_1.01,0.057062399387359616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,2,1,power_law_1.01,0.12972160577774047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,2,1,power_law_1.01,0.0839680016040802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,2,1,power_law_1.01,0.13012479543685912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,2,1,power_law_1.01,0.08997759819030762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,2,1,power_law_1.01,0.1301632046699524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,2,1,power_law_1.01,0.09169920086860657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,2,1,power_law_1.01,0.13156479597091675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,2,1,power_law_1.01,0.0954688012599945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,2,1,power_law_1.01,0.1351423978805542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,2,1,power_law_1.01,0.09547520279884339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,2,1,power_law_1.01,0.1391808032989502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,2,1,power_law_1.01,0.0961023986339569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,2,1,power_law_1.01,0.15720319747924805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,2,1,power_law_1.01,0.09674879908561707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,2,1,power_law_1.01,0.1862912058830261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,2,1,power_law_1.01,0.09748479723930359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,2,1,power_law_1.01,0.20936319828033448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,2,1,power_law_1.01,0.09958400130271912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,2,1,power_law_1.01,0.24036478996276855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,2,1,power_law_1.01,0.10248960256576538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,2,1,power_law_1.01,0.27845120429992676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,2,1,power_law_1.01,0.3454463958740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,2,1,power_law_1.01,0.4174655914306641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,2,1,power_law_1.01,0.5673535823822021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,2,power_law_1.2,0.9788928031921387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,2,1,power_law_1.01,0.7160575866699219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,2,power_law_1.2,1.211616039276123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,2,1,power_law_1.01,1.0161791801452638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,2,1,power_law_1.01,0.10703359842300415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,2,1,power_law_1.01,0.1187391996383667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,2,1,power_law_1.01,0.1429695963859558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,2,1,power_law_1.01,1.3237631797790528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,2,1,power_law_1.01,0.16551680564880372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,2,1,power_law_1.01,0.19000320434570311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,2,1,power_law_1.01,0.22602880001068115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,2,1,power_law_1.01,0.28755199909210205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,2,1,power_law_1.01,0.3468031883239746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,2,1,power_law_1.01,0.4798592090606689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,2,1,power_law_1.01,0.6078144073486328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,2,1,power_law_1.01,0.02497279942035675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,2,1,power_law_1.01,0.8660927772521972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,2,1,power_law_1.01,0.026208001375198364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,2,1,power_law_1.01,0.030643200874328612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,2,power_law_1.2,1.3312447547912598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,2,1,power_law_1.01,0.03335680067539215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,2,1,power_law_1.01,1.1262720108032227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,2,1,power_law_1.01,0.0411327987909317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,2,1,power_law_1.2,0.021376000344753267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,2,1,power_law_1.01,0.05123839974403381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,2,1,power_law_1.2,0.031769600510597226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,2,1,power_law_1.01,0.055270397663116456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,2,1,power_law_1.2,0.044460800290107724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,2,1,power_law_1.01,0.05841919779777527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,2,1,power_law_1.2,0.06517120003700257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,2,1,power_law_1.01,0.05913599729537964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,2,1,power_law_1.2,0.07461119890213012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,2,1,power_law_1.01,0.058931201696395874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,2,1,power_law_1.01,0.028249600529670717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,2,1,power_law_1.2,0.11474560499191284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,2,1,power_law_1.01,0.061843198537826535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,2,1,power_law_1.2,0.12905600070953369
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,2,1,power_law_1.01,0.06467199921607972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,2,1,power_law_1.2,0.13235199451446533
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,2,1,power_law_1.01,0.06689919829368592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,2,1,power_law_1.2,0.13730560541152953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,2,1,power_law_1.01,0.073471999168396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,2,1,power_law_1.2,0.13859200477600098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,2,1,power_law_1.01,0.0676800012588501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,2,1,power_law_1.2,0.12956160306930542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,2,1,power_law_1.01,0.07086079716682434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,2,1,power_law_1.2,0.12992639541625978
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,2,1,power_law_1.01,0.07843199968338013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,2,1,power_law_1.2,0.1314560055732727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,2,1,power_law_1.01,0.08275200128555298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,2,1,power_law_1.2,0.1326848030090332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,2,1,power_law_1.2,0.13725440502166747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,2,1,power_law_1.2,0.14869120121002197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,2,1,power_law_1.2,0.1605247974395752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,2,power_law_1.2,0.3324415922164917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,2,1,power_law_1.2,0.20254719257354736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,2,1,power_law_1.2,0.2151103973388672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,2,1,power_law_1.2,0.24082560539245607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,2,1,power_law_1.2,0.28178560733795166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,2,1,power_law_1.01,0.09376000165939331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,2,1,power_law_1.2,0.35000319480895997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,2,1,power_law_1.01,0.10295039415359497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,2,1,power_law_1.01,0.11998720169067383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,2,1,power_law_1.01,0.1679103970527649
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,2,1,power_law_1.01,0.2038655996322632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,2,1,power_law_1.01,0.2701632022857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,2,1,power_law_1.01,0.33844480514526365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,2,1,power_law_1.01,0.4655871868133545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,2,1,power_law_1.2,0.4295231819152832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,2,1,power_law_1.01,0.608243179321289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,2,1,power_law_1.2,0.021766400337219237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,2,1,power_law_1.2,0.027302399277687073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,2,1,balanced,0.42131733894348145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,2,1,power_law_1.2,0.033011201024055484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,2,1,balanced,0.5499039888381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,2,1,power_law_1.2,0.045817598700523376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,2,1,balanced,0.7853653430938721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,2,1,power_law_1.2,0.05397760272026062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,2,1,power_law_1.2,0.0834496021270752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,2,1,power_law_1.2,0.08960639834403991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,2,power_law_1.2,1.083193588256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,2,1,power_law_1.2,0.0923583984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,2,1,power_law_1.2,0.09651200175285339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,2,1,power_law_1.2,0.0953984022140503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,2,1,power_law_1.2,0.09486079812049866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,2,1,power_law_1.2,0.09694719910621644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,2,1,power_law_1.2,0.09701120257377624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,2,1,power_law_1.2,0.09917439818382263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,2,1,balanced,1.032639980316162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,2,1,power_law_1.2,0.10363520383834839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,2,1,power_law_1.2,0.10753920078277587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,2,1,power_law_1.01,0.032179200649261476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,2,1,balanced,1.4993386268615723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,2,1,power_law_1.2,0.11956479549407958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,2,1,power_law_1.01,0.04008319973945618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,2,1,power_law_1.2,0.15181440114974976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,2,1,power_law_1.01,0.053990399837493895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,2,1,power_law_1.2,0.16700799465179444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,2,1,balanced,2.000410715738932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,2,1,power_law_1.2,0.5712192058563232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,2,1,power_law_1.2,0.1965183973312378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,2,1,power_law_1.2,0.22935678958892822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,2,1,power_law_1.2,0.28976640701293943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,2,1,power_law_1.2,0.02852480113506317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,2,1,power_law_1.2,0.032716798782348636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,2,1,power_law_1.2,0.04017919898033142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,2,1,power_law_1.2,0.3582592010498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,2,1,power_law_1.2,0.05186560153961182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,2,1,power_law_1.2,0.05800319910049438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,2,power_law_1.2,0.4771903991699219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,2,1,power_law_1.2,0.08828799724578858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,2,1,power_law_1.2,0.09165440201759338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,2,1,power_law_1.2,0.09496960043907166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,2,1,power_law_1.2,0.09528319835662842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,2,1,power_law_1.2,0.09758080244064331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,2,1,power_law_1.2,0.10563199520111084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,2,1,power_law_1.2,0.11234560012817382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,2,1,power_law_1.2,0.11598080396652222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,2,1,power_law_1.2,0.11679359674453735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,2,1,power_law_1.2,0.12691839933395385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,2,1,power_law_1.2,0.1414463996887207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,2,1,power_law_1.2,0.18261760473251343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,2,1,power_law_1.2,0.20355839729309083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,2,1,power_law_1.2,0.23798398971557616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,2,1,power_law_1.2,0.2900160074234009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,2,1,power_law_1.2,0.35068159103393554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,2,1,power_law_1.2,0.4853824138641357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,2,1,power_law_1.01,0.061337602138519284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,2,1,power_law_1.2,0.7253248214721679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,2,1,power_law_1.2,0.4660223960876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,2,1,power_law_1.2,0.5835968017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,2,power_law_1.2,1.3962368011474608
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,2,1,power_law_1.2,0.02222079932689667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,2,1,power_law_1.2,0.023827199637889863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,2,1,power_law_1.2,0.8142720222473144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,2,1,power_law_1.2,0.027622398734092713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,2,1,power_law_1.2,0.030675199627876282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,2,1,power_law_1.2,0.03769600093364715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,2,1,power_law_1.2,0.05090559720993042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,2,1,power_law_1.2,0.054688000679016115
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,2,1,power_law_1.2,0.05690240263938904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,2,1,power_law_1.2,0.05860480070114136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,2,1,power_law_1.2,0.059462398290634155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,2,1,power_law_1.2,0.06352639794349671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,2,1,power_law_1.2,0.06474239826202392
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,2,1,power_law_1.2,0.06676480174064636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,2,1,power_law_1.2,0.07382400035858154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,2,1,power_law_1.01,0.08951039910316468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,2,1,power_law_1.2,0.06839680075645446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,2,1,power_law_1.01,0.0893119990825653
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,2,1,power_law_1.2,0.07171840071678162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,2,1,power_law_1.01,0.0923583984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,2,1,power_law_1.2,0.07871360182762147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,2,1,power_law_1.01,0.09315199851989746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,2,1,power_law_1.2,0.08495360016822814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,2,1,power_law_1.01,0.09365760087966919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,2,1,power_law_1.01,0.1007423996925354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,2,1,power_law_1.01,0.11136000156402588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,2,1,power_law_1.01,0.1155392050743103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,2,1,power_law_1.01,0.11578240394592285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,2,1,power_law_1.01,0.126528000831604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,2,1,power_law_1.01,0.13865599632263184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,2,1,power_law_1.2,1.0278976440429688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,2,1,power_law_1.2,0.6145023822784423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,2,1,power_law_1.2,0.09500799775123596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,2,1,power_law_1.2,0.10406399965286255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,2,1,power_law_1.2,0.12272640466690063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,2,1,power_law_1.2,0.16963839530944824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,2,1,power_law_1.2,0.20611200332641602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,2,power_law_1.2,1.9726015090942384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,2,1,power_law_1.2,0.27370240688323977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,2,1,power_law_1.2,0.874118423461914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,2,power_law_1.2,2.467532730102539
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,2,1,power_law_1.2,0.3411648035049438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,2,balanced,0.055786664287249245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,2,balanced,0.08037866652011871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,2,balanced,0.1291146675745646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,2,balanced,0.21754133701324463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,2,balanced,0.38198399543762207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,2,balanced,0.38577067852020264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,2,balanced,0.4137386480967204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,2,balanced,0.4137226740519206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,2,balanced,0.43013866742451984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,2,balanced,0.4331839879353841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,2,balanced,0.425653338432312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,2,balanced,0.42742399374643963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,2,1,power_law_1.01,0.18012800216674804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,2,1,power_law_1.01,0.19896960258483887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,2,1,power_law_1.01,0.23439359664916992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,2,1,power_law_1.01,0.29009280204772947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,2,1,power_law_1.01,0.34688639640808105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,2,balanced,0.43294934431711835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,2,balanced,0.43832000096638996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,2,1,power_law_1.2,1.0527487754821778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,2,balanced,0.4457706610361735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,2,balanced,0.45323201020558673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,2,balanced,0.46722666422526044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,2,balanced,0.5620160102844238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,2,balanced,0.5842933257420858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,2,balanced,0.6304159959157308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,2,balanced,0.6780266761779785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,2,balanced,0.9964106877644857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,2,balanced,1.1096373399098713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,2,balanced,1.5598079363505046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,2,balanced,2.011125405629476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,2,balanced,2.9408000310262046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,2,balanced,3.908426602681478
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,2,power_law_1.2,0.6300032138824463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,2,balanced,0.045552000403404236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,2,balanced,0.06634666522343953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,2,balanced,0.08927466471989949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,2,balanced,0.1395039955774943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,2,balanced,0.2266133427619934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,2,balanced,0.2300800085067749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,2,balanced,0.2640533248583476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,2,balanced,0.2661813298861186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,2,balanced,0.25148266553878784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,2,balanced,0.25387734174728394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,2,balanced,0.25338133176167804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,2,balanced,0.2534613410631816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,2,balanced,0.25935999552408856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,2,balanced,0.26687999566396076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,2,balanced,0.27405865987141925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,2,balanced,0.28199466069539386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,2,balanced,0.2949066758155823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,2,1,power_law_1.01,0.469817590713501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,2,balanced,0.35809600353240967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,2,balanced,0.3834506670633952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,2,1,power_law_1.01,0.5836991786956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,2,balanced,0.4340426524480184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,2,balanced,0.4855360190073649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,2,1,power_law_1.01,0.8140864372253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,2,1,power_law_1.01,1.0502655982971192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,2,1,power_law_1.01,1.5114175796508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,2,balanced,0.6998079617818197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,2,1,power_law_1.01,1.9762687683105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,2,balanced,0.8042026360829672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,2,balanced,1.1305973529815674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,2,balanced,1.455407937367757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,2,balanced,0.053632001082102455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,2,balanced,0.06855999926726024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,2,balanced,0.090037335952123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,2,balanced,0.14933866262435913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,2,balanced,0.23676800727844238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,2,balanced,0.24146666129430136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,2,balanced,2.132810592651367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,2,balanced,0.24388800064722696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,2,balanced,0.2456159989039103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,2,balanced,0.3980426788330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,2,balanced,0.39977065722147626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,2,balanced,0.4052480061848958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,2,balanced,2.8242241541544595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,2,balanced,0.4123893181482951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,2,balanced,0.418938676516215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,2,power_law_1.01,0.07553279995918274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,2,balanced,0.42668267091115314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,2,balanced,0.4372853438059489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,2,power_law_1.01,0.10669440031051636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,2,balanced,0.44626132647196454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,2,power_law_1.01,0.14451199769973755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,2,balanced,0.4673173427581787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,2,power_law_1.01,0.19638400077819823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,2,power_law_1.01,0.23374719619750978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,2,power_law_1.01,0.3393280029296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,2,balanced,0.02887466549873352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,2,balanced,0.03150933235883713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,2,balanced,0.054085334142049156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,2,power_law_1.01,0.3718208074569702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,2,power_law_1.01,0.3753664016723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,2,power_law_1.01,0.4174464225769043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,2,1,power_law_1.2,0.471724796295166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,2,balanced,0.5018239816029867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,2,balanced,0.5374240080515543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,2,balanced,0.8437279860178629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,2,balanced,0.9194719791412354
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,2,balanced,0.08029866715272267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,2,balanced,0.1242026686668396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,2,balanced,0.13004799683888754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,2,balanced,1.3196106751759846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,2,balanced,1.716383934020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,2,power_law_1.01,0.4244351863861084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,2,power_law_1.01,0.4088768005371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,2,power_law_1.01,0.41510400772094724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,2,power_law_1.01,0.42574081420898435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,2,power_law_1.01,0.42970881462097166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,2,balanced,0.1320853332678477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,2,power_law_1.01,0.44535040855407715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,2,balanced,0.13385599851608276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,2,balanced,0.13586666186650595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,2,balanced,0.1379680037498474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,2,power_law_1.01,0.44931840896606445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,2,balanced,0.14139200250307718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,2,balanced,0.13793599605560303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,2,balanced,0.14217600226402283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,2,power_law_1.01,0.4886591911315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,2,balanced,0.14482133587201437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,2,balanced,0.14628266294797262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,2,balanced,0.14949867129325867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,2,balanced,0.156031996011734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,2,power_law_1.01,0.5916672229766846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,2,balanced,2.5052639643351235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,2,balanced,0.16565866271654764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,2,1,power_law_1.2,1.5155263900756837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,2,power_law_1.01,0.6346047878265381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,2,power_law_1.01,0.7552127838134766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,2,balanced,3.29586124420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,2,power_law_1.01,0.8778752326965332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,2,power_law_1.01,1.1159680366516114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,2,balanced,4.8258771896362305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,2,balanced,0.1718399922053019
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,2,balanced,0.2347093423207601
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,2,balanced,0.20350400606791177
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,2,balanced,0.3787573178609212
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,2,balanced,0.31616532802581787
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,2,balanced,0.44682133197784424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,2,balanced,0.5782933235168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,2,1,power_law_1.2,1.3348608016967773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,2,balanced,0.8392586708068848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,2,power_law_1.01,0.05621119737625122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,2,balanced,6.388720194498698
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,2,balanced,1.0997172991434734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,2,power_law_1.01,0.06869120001792908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,2,power_law_1.01,0.09418240189552307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,2,power_law_1.01,1.3342399597167969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,2,power_law_1.01,0.1250432014465332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,2,power_law_1.01,0.06572800278663635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,2,power_law_1.01,0.1435968041419983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,2,power_law_1.01,0.07288960218429566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,2,power_law_1.01,1.7916351318359376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,2,power_law_1.01,0.1987264037132263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,2,power_law_1.01,0.2388159990310669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,2,power_law_1.01,0.24351999759674073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,2,power_law_1.01,0.23266561031341554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,2,power_law_1.01,2.3035263061523437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,2,power_law_1.01,0.23444480895996095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,2,power_law_1.01,0.2413696050643921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,2,power_law_1.01,0.24406399726867675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,2,power_law_1.01,0.2527935981750488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,2,power_law_1.01,0.10599679946899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,2,power_law_1.01,0.2579456090927124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,2,power_law_1.01,3.2916927337646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,2,power_law_1.01,0.12574080228805543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,2,power_law_1.01,0.2679487943649292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,2,power_law_1.01,0.15773439407348633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,2,power_law_1.01,0.28218879699707033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,2,power_law_1.01,0.22248320579528807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,2,power_law_1.01,0.30018560886383056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,2,power_law_1.01,0.030508801341056824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,2,power_law_1.01,0.2188096046447754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,2,power_law_1.01,0.04801279902458191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,2,power_law_1.01,0.37475199699401857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,2,power_law_1.01,0.23773438930511476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,2,power_law_1.01,0.0571008026599884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,2,power_law_1.01,0.4136832237243652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,2,power_law_1.01,0.34771840572357177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,2,power_law_1.01,4.407257461547852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,2,power_law_1.01,0.07336320281028748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,2,power_law_1.01,0.36412160396575927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,2,power_law_1.01,0.5001599788665771
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,2,power_law_1.01,0.08600320219993592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,2,power_law_1.2,0.08128640055656433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,2,power_law_1.01,0.37034881114959717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,2,power_law_1.01,0.5822591781616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,2,power_law_1.01,0.11552640199661254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,2,power_law_1.2,0.10040960311889649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,2,power_law_1.01,0.3790143966674805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,2,1,power_law_1.2,1.1468671798706054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,2,power_law_1.01,0.7584896087646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,2,power_law_1.01,0.40712318420410154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,2,power_law_1.2,0.05441920161247253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,2,power_law_1.01,0.4312128067016602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,2,power_law_1.2,0.06789759993553161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,2,power_law_1.01,0.9336959838867187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,2,power_law_1.2,0.0986624002456665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,2,power_law_1.2,0.12334719896316529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,2,power_law_1.01,1.2913984298706054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,2,power_law_1.2,0.13628159761428832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,2,power_law_1.2,0.2013887882232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,2,power_law_1.2,0.23714559078216552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,2,power_law_1.01,1.6350143432617188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,2,power_law_1.01,0.12028160095214843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,2,1,power_law_1.2,0.6211647987365723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,2,power_law_1.2,0.24443519115447998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,2,power_law_1.01,0.12348159551620483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,2,power_law_1.2,0.22970240116119384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,2,power_law_1.2,0.24240639209747314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,2,power_law_1.01,0.45205121040344237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,2,power_law_1.2,0.06556159853935242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,2,power_law_1.2,0.24406399726867675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,2,power_law_1.2,0.07257599830627441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,2,power_law_1.01,0.4742464065551758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,2,power_law_1.2,0.25118720531463623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,2,power_law_1.2,0.09838079810142517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,2,power_law_1.01,0.5201087951660156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,2,power_law_1.2,0.25138559341430666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,2,power_law_1.2,0.13118079900741578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,2,power_law_1.2,0.2637439966201782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,2,power_law_1.01,0.6199872016906738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,2,power_law_1.2,0.15685759782791137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,2,power_law_1.2,0.14058879613876343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,2,power_law_1.2,0.2762367963790894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,2,power_law_1.2,0.22269439697265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,2,power_law_1.2,0.17889280319213868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,2,power_law_1.01,0.7191936016082764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,2,power_law_1.2,0.2918720006942749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,2,power_law_1.2,0.2339776039123535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,2,power_law_1.01,0.1288831949234009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,2,power_law_1.2,0.219596791267395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,2,power_law_1.2,0.30738561153411864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,2,power_law_1.2,0.23557119369506835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,2,power_law_1.01,0.13219200372695922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,2,power_law_1.2,0.3436032056808472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,2,power_law_1.01,0.9246208190917968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,2,power_law_1.2,0.3819648027420044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,2,power_law_1.01,0.1404736042022705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,2,power_law_1.2,0.3739392042160034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,2,power_law_1.2,0.35865600109100343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,2,power_law_1.01,0.13343360424041747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,2,power_law_1.2,0.422214412689209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,2,power_law_1.2,0.379366397857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,2,power_law_1.2,0.38195199966430665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,2,power_law_1.01,1.1534655570983887
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,2,power_law_1.01,0.13965439796447754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,2,power_law_1.2,0.5036736011505127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,2,power_law_1.2,0.38618879318237304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,2,power_law_1.2,0.4199552059173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,2,power_law_1.01,0.14840960502624512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,2,power_law_1.2,0.3909375905990601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,2,power_law_1.2,0.41635842323303224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,2,power_law_1.2,0.5887296199798584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,2,power_law_1.01,0.14762879610061647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,2,power_law_1.01,1.5865983963012695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,2,power_law_1.2,0.41626877784729005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,2,power_law_1.2,0.40610561370849607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,2,power_law_1.2,0.7765312194824219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,2,power_law_1.2,0.4388415813446045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,2,power_law_1.2,0.42030720710754393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,2,power_law_1.01,2.0166528701782225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,2,power_law_1.2,0.41923837661743163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,2,power_law_1.2,0.43944320678710935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,2,power_law_1.2,0.44849281311035155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,2,power_law_1.2,0.4707007884979248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,2,power_law_1.01,2.750163269042969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,2,power_law_1.2,0.49845118522644044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,2,power_law_1.01,0.15185920000076295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,2,power_law_1.2,0.6042496204376221
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,2,power_law_1.01,0.16535680294036864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,2,power_law_1.2,0.9445247650146484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,2,power_law_1.01,0.17626880407333373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,2,power_law_1.2,0.4500864028930664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,2,power_law_1.2,0.6569983959197998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,2,power_law_1.01,0.19125759601593018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,2,power_law_1.2,0.48321919441223143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,2,power_law_1.01,3.786886215209961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,2,power_law_1.2,0.7819392204284668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,2,power_law_1.01,0.25597438812255857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,2,power_law_1.2,0.5422783851623535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,2,power_law_1.01,0.2633023977279663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,2,power_law_1.2,0.6503551959991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,2,power_law_1.01,0.3341952085494995
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,2,power_law_1.01,0.3913599967956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,2,power_law_1.2,1.3223872184753418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,2,power_law_1.01,0.541548776626587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,2,power_law_1.2,1.6429567337036133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,2,power_law_1.01,5.092979049682617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,2,power_law_1.2,0.9190848350524903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,2,power_law_1.01,0.6549248218536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,2,power_law_1.01,0.9872639656066895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,2,power_law_1.2,0.7578303813934326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,2,power_law_1.01,1.5103039741516113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,2,power_law_1.2,0.9819583892822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,2,power_law_1.01,6.536153411865234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,2,power_law_1.2,1.1228416442871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,2,1,balanced,0.03344533344109853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,2,1,balanced,0.051776001850763954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,2,1,balanced,0.07856533428033192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,2,1,balanced,0.13581867019335428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,2,power_law_1.2,1.6722240447998047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,2,1,balanced,0.2257759968439738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,2,1,balanced,0.22887466351191202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,2,power_law_1.2,1.137382411956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,2,1,balanced,0.26375999053319293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,2,1,balanced,0.2643466591835022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,2,1,balanced,0.24843200047810873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,2,power_law_1.2,1.6116928100585937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,2,1,balanced,0.24882133801778158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,2,1,balanced,0.2432639996210734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,2,1,balanced,0.2453333338101705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,2,power_law_1.01,2.3019903182983397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,2,1,balanced,0.24764267603556314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,2,power_law_1.2,0.030585598945617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,2,power_law_1.2,0.04519680142402649
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,2,power_law_1.2,0.05820159912109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,2,power_law_1.2,0.06855040192604064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,2,power_law_1.01,3.0872064590454102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,2,power_law_1.2,0.0786624014377594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,2,power_law_1.2,0.11304960250854493
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,2,power_law_1.2,0.12205439805984497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,2,power_law_1.2,0.12495360374450684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,2,1,balanced,0.2525920073191325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,2,power_law_1.2,0.13050880432128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,2,1,balanced,0.25861332813898724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,2,1,balanced,0.2635519901911418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,2,power_law_1.2,0.13038079738616942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,2,power_law_1.2,0.14371199607849122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,2,power_law_1.2,0.13693439960479736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,2,power_law_1.2,1.4263808250427246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,2,power_law_1.2,0.1397312045097351
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,2,power_law_1.2,0.15338879823684692
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,2,power_law_1.2,0.15121920108795167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,2,power_law_1.2,0.15515520572662353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,2,1,balanced,0.04734933376312256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,2,power_law_1.2,0.17137279510498046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,2,1,balanced,0.2723413308461507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,2,power_law_1.2,0.18069759607315064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,2,1,balanced,0.3428959846496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,2,1,balanced,0.360207994778951
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,2,power_law_1.2,0.2047935962677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,2,1,balanced,0.3970133463541667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,2,1,balanced,0.4336106777191162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,2,power_law_1.2,0.2327552080154419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,2,power_law_1.2,1.9571392059326171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,2,1,balanced,0.6421546538670858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,2,power_law_1.2,0.26689920425415037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,2,1,balanced,0.7159466743469238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,2,power_law_1.2,0.3479423999786377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,2,1,balanced,1.0101333459218342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,2,1,balanced,1.2965760231018066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,2,1,balanced,1.8997279802958171
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,2,power_law_1.2,0.4101439952850342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,2,1,balanced,2.4908053080240884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,2,power_law_1.2,2.8618751525878907
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,2,power_law_1.2,0.5968255996704102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,2,1,power_law_1.2,1.9783424377441405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,2,1,balanced,0.07382399837176006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,2,1,balanced,0.028597332537174225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,2,1,balanced,0.03549866626660029
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,2,1,balanced,0.05413866539796194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,2,1,balanced,0.0786240001519521
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,2,1,balanced,0.1254026691118876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,2,1,balanced,0.12820266683896384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,2,1,balanced,0.13132799665133157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,2,1,balanced,0.13593066732088724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,2,power_law_1.2,1.8848384857177733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,2,1,balanced,0.1367093324661255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,2,1,balanced,0.13983999689420065
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,2,1,balanced,0.14282666643460593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,2,1,balanced,0.1612320045630137
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,2,1,balanced,0.16478400429089865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,2,1,balanced,0.17088532447814941
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,2,1,balanced,0.1688906749089559
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,2,1,balanced,0.1729653278986613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,2,1,balanced,0.04228266576925913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,2,1,balanced,0.18125865856806436
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,2,1,balanced,0.19532267252604166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,2,1,balanced,0.20626133680343628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,2,1,balanced,0.28018667300542194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,2,1,balanced,0.2596319913864136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,2,1,balanced,0.5625386635462443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,2,1,balanced,0.34397868315378827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,2,power_law_1.2,2.3907264709472655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,2,1,balanced,0.48955198129018146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,2,1,balanced,0.6284106572469076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,2,1,balanced,0.9014346599578857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,2,1,balanced,1.1830560366312664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,2,1,power_law_1.01,0.031564798951148984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,2,1,power_law_1.01,0.04922240078449249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,2,1,power_law_1.01,0.07553279995918274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,2,1,power_law_1.01,0.12125439643859863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,2,1,power_law_1.01,0.13615360260009765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,2,1,power_law_1.01,0.19465600252151488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,2,1,power_law_1.01,0.2373120069503784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,2,power_law_1.2,3.7005569458007814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,2,1,power_law_1.01,0.2416896104812622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,2,1,balanced,0.12363200386365254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,2,1,power_law_1.01,0.23423359394073487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,2,1,power_law_1.01,0.04559360146522522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,2,1,power_law_1.01,0.23633279800415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,2,1,power_law_1.01,0.2334399938583374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,2,1,power_law_1.01,0.23552000522613525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,2,1,power_law_1.01,0.24087679386138916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,2,1,power_law_1.01,0.24954240322113036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,2,1,power_law_1.01,0.25516159534454347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,2,1,power_law_1.01,0.2642944097518921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,2,1,power_law_1.01,0.2832767963409424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,2,1,power_law_1.01,0.3518143892288208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,2,1,power_law_1.01,0.39315199851989746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,2,1,power_law_1.01,0.4616191864013672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,2,1,power_law_1.01,0.5349760055541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,2,power_law_1.2,2.2693952560424804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,2,1,power_law_1.01,0.6733312129974365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,2,1,power_law_1.01,0.8201536178588867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,2,1,power_law_1.01,1.1236224174499512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,2,1,power_law_1.01,1.4240320205688477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,2,1,power_law_1.01,2.012972831726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,2,1,balanced,0.05811200042565664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,2,1,balanced,0.0824480007092158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,2,1,balanced,0.14564800262451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,2,1,balanced,0.20928533871968588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,2,1,power_law_1.01,0.07037439942359924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,2,1,power_law_1.01,2.6229183197021486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,2,power_law_1.2,3.472281646728516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,2,1,power_law_1.01,0.11592320203781128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,2,1,balanced,0.3760106563568115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,2,1,balanced,0.3791733185450236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,2,1,balanced,0.2351306676864624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,2,power_law_1.2,0.7119232177734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,2,1,power_law_1.01,0.04229759871959686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,2,power_law_1.2,4.539519882202148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,2,1,power_law_1.01,0.15994240045547486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,2,1,balanced,0.3821440140406291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,2,1,balanced,0.38469866911570233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,2,1,balanced,0.38504000504811603
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,2,1,power_law_1.01,0.02863360047340393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,2,1,power_law_1.01,0.03617919981479645
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,2,1,power_law_1.01,0.054041600227355956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,2,1,power_law_1.01,0.06665599942207337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,2,1,power_law_1.01,0.08305919766426087
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,2,1,power_law_1.01,0.1142016053199768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,2,1,power_law_1.01,0.1193727970123291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,2,1,power_law_1.01,0.12513920068740844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,2,1,power_law_1.01,0.12993279695510865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,2,1,power_law_1.01,0.13337600231170654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,2,1,power_law_1.01,0.14266879558563234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,2,1,power_law_1.01,0.15386879444122314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,2,1,power_law_1.01,0.16244479417800903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,2,1,power_law_1.01,0.18468480110168456
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,2,1,power_law_1.01,0.1637120008468628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,2,1,power_law_1.01,0.17434879541397094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,2,1,power_law_1.01,0.19848320484161378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,2,1,power_law_1.01,0.20910720825195311
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,2,1,power_law_1.01,0.23269119262695312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,2,1,power_law_1.01,0.24966399669647216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,2,1,balanced,0.38362665971120197
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,2,power_law_1.2,0.9586175918579102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,2,1,power_law_1.01,0.29231359958648684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,2,1,power_law_1.01,0.21748480796813965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,2,1,balanced,0.25357866287231445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,2,1,balanced,0.23880000909169516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,2,1,power_law_1.01,0.32266879081726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,2,1,power_law_1.01,0.3202944040298462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,2,1,balanced,0.2405280073483785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,2,1,balanced,0.24143999814987183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,2,1,power_law_1.01,0.3912384033203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,2,1,balanced,0.24226667483647665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,2,1,balanced,0.24427199363708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,2,1,balanced,0.24767466386159262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,2,1,power_law_1.01,0.535475206375122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,2,1,balanced,0.25249600410461426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,2,1,balanced,0.25685866673787433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,2,1,balanced,0.3577066659927368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,2,power_law_1.2,5.105068969726562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,2,1,power_law_1.01,0.6750336170196534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,2,1,balanced,0.36321067810058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,2,1,power_law_1.01,0.057158398628234866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,2,1,power_law_1.01,0.9484479904174805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,2,1,power_law_1.01,0.07957760095596314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,2,1,power_law_1.01,0.11598720550537109
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,2,1,power_law_1.01,1.3918208122253417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,2,1,balanced,0.38708798090616864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,2,1,balanced,0.39079999923706055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,2,1,balanced,0.3927626609802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,2,1,power_law_1.01,0.32785279750823976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,2,1,power_law_1.01,0.3473344087600708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,2,1,power_law_1.01,0.3589632034301758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,2,1,power_law_1.2,0.04602240025997162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,2,1,balanced,0.394869327545166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,2,1,balanced,0.5092426538467407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,2,1,power_law_1.01,0.15147520303726197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,2,1,balanced,0.5129173199335734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,2,1,balanced,0.4481920003890991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,2,1,balanced,0.46832001209259033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,2,1,balanced,0.49423468112945557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,2,1,power_law_1.2,0.07033600211143494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,2,1,power_law_1.01,0.36624000072479246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,2,1,balanced,0.4453546603520711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,2,1,balanced,0.4726453224817912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,2,1,balanced,0.5768053531646729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,2,1,balanced,0.6153119802474976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,2,1,power_law_1.2,0.11350400447845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,2,1,power_law_1.01,0.3809216022491455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,2,1,power_law_1.01,0.39864959716796877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,2,1,power_law_1.01,0.4156735897064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,2,1,power_law_1.01,0.2133631944656372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,2,1,power_law_1.01,0.2215872049331665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,2,1,power_law_1.01,0.22610559463500976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,2,1,power_law_1.01,0.22860159873962402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,2,1,power_law_1.01,0.22882559299468994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,2,1,power_law_1.01,0.24056320190429686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,2,1,power_law_1.01,0.26358399391174314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,2,1,power_law_1.01,0.27618560791015623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,2,1,power_law_1.01,0.30785279273986815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,2,1,power_law_1.01,0.3370368003845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,2,1,power_law_1.01,0.38056960105896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,2,1,power_law_1.2,0.15731199979782104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,2,power_law_1.2,3.0938047409057616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,2,1,power_law_1.2,0.034041601419448855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,2,1,power_law_1.2,0.04925439953804016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,2,1,power_law_1.2,0.06959999799728393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,2,1,power_law_1.2,0.10414079427719117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,2,1,power_law_1.2,0.12929919958114625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,2,1,power_law_1.01,0.4417088031768799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,2,1,power_law_1.2,0.19496320486068724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,2,1,power_law_1.2,0.24360320568084717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,2,1,power_law_1.01,0.46935038566589354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,2,1,power_law_1.2,0.24487679004669188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,2,1,power_law_1.01,0.5219520092010498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,2,1,power_law_1.2,0.2349247932434082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,2,1,power_law_1.01,0.4672192096710205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,2,1,power_law_1.2,0.23827199935913085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,2,1,power_law_1.01,0.5189375877380371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,2,1,power_law_1.2,0.23785600662231446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,2,1,power_law_1.2,0.2408384084701538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,2,1,power_law_1.01,0.5649600028991699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,2,1,power_law_1.2,0.24031999111175537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,2,1,power_law_1.2,0.24938879013061524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,2,1,power_law_1.2,0.2579008102416992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,2,1,power_law_1.2,0.27509119510650637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,2,1,power_law_1.2,0.2903680086135864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,2,1,power_law_1.2,0.3615295886993408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,2,1,power_law_1.2,0.3951551914215088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,2,1,power_law_1.2,0.4627200126647949
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,2,power_law_1.2,1.560313606262207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,2,1,power_law_1.2,0.5418303966522217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,2,1,power_law_1.2,0.040761598944664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,2,1,power_law_1.2,0.6818431854248047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,2,1,power_law_1.2,0.05689600110054016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,2,1,power_law_1.2,0.18933759927749633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,2,1,power_law_1.2,0.07646719813346863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,2,1,power_law_1.2,0.31746559143066405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,2,1,power_law_1.2,0.1028223991394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,2,1,power_law_1.2,0.13955199718475342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,2,1,power_law_1.2,0.2092223882675171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,2,1,power_law_1.2,0.22429440021514893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,2,1,power_law_1.2,0.22906239032745362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,2,1,power_law_1.2,0.22851839065551757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,2,1,balanced,0.5018560091654459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,2,1,power_law_1.2,0.23788158893585204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,2,1,balanced,0.7962559858957926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,2,1,power_law_1.2,0.2486720085144043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,2,1,balanced,0.8480479717254639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,2,1,power_law_1.2,0.2634239912033081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,2,1,power_law_1.2,0.2760447978973389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,2,1,power_law_1.2,0.8316415786743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,2,1,balanced,1.1971466541290283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,2,1,power_law_1.2,0.3101632118225098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,2,1,balanced,1.5686720212300618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,2,1,power_law_1.2,0.34244480133056643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,2,1,balanced,0.8183573087056478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,2,1,power_law_1.2,0.38535680770874026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,2,1,balanced,0.8886240323384603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,2,1,power_law_1.2,0.49543042182922364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,2,1,balanced,2.2600533167521157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,2,1,balanced,1.4115573565165203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,2,1,power_law_1.2,0.5813695907592773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,2,1,balanced,1.5678240458170574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,2,1,power_law_1.2,0.6703680038452149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,2,1,balanced,3.0174080530802407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,2,1,power_law_1.2,0.8284735679626465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,2,1,balanced,2.3331680297851562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,2,1,power_law_1.2,1.003609561920166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,2,1,balanced,4.325434684753418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,2,1,balanced,3.045210520426432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,2,1,power_law_1.2,1.3503616333007813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,2,1,power_law_1.2,1.133471965789795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,2,1,power_law_1.2,1.66125431060791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,2,1,balanced,5.84771728515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,2,1,power_law_1.2,0.3401472091674805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,2,balanced,0.08467732866605122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,2,1,power_law_1.2,0.358515191078186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,2,1,power_law_1.2,0.37377281188964845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,2,1,power_law_1.2,0.02714880108833313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,2,balanced,0.14601600170135498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,2,1,power_law_1.2,0.0369024008512497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,2,balanced,0.24901866912841797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,2,balanced,0.459114670753479
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,2,1,power_law_1.2,0.049446401000022885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,2,balanced,0.8778186639149984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,2,1,power_law_1.2,0.06251519918441772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,2,balanced,1.0943893591562908
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,2,1,power_law_1.2,0.07972480058670044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,2,1,power_law_1.01,0.6450111865997314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,2,1,power_law_1.2,0.11325440406799317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,2,balanced,1.1288106441497803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,2,1,power_law_1.01,0.7328256130218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,2,balanced,1.1308053334554036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,2,balanced,1.28165864944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,2,1,power_law_1.01,0.4911168098449707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,2,balanced,1.2870773474375408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,2,1,power_law_1.01,0.565990400314331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,2,balanced,1.232682704925537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,2,balanced,1.2377493381500244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,2,1,power_law_1.2,2.3089664459228514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,2,balanced,1.2441173394521077
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,2,1,power_law_1.2,0.12037119865417481
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,2,1,power_law_1.2,0.12643840312957763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,2,1,power_law_1.2,0.1301375985145569
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,2,1,power_law_1.2,0.13593599796295167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,2,1,power_law_1.2,0.14424959421157837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,2,1,power_law_1.2,0.15306880474090576
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,2,1,power_law_1.2,0.16711039543151857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,2,1,power_law_1.2,0.18685439825057984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,2,1,power_law_1.2,0.3675519943237305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,2,1,power_law_1.2,0.16521600484848023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,2,balanced,1.2619360287984211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,2,1,power_law_1.2,0.17974400520324707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,2,1,power_law_1.2,0.20249600410461427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,2,balanced,1.2579200267791748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,2,1,power_law_1.2,0.20599040985107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,2,balanced,1.269477367401123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,2,1,power_law_1.2,1.4382783889770507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,2,1,power_law_1.2,0.23983359336853027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,2,balanced,1.2912906805674236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,2,1,power_law_1.2,0.2651711940765381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,2,1,power_law_1.2,0.29562880992889407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,2,balanced,1.4890987078348796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,2,1,power_law_1.2,0.3367487907409668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,2,balanced,1.5328000386555989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,2,balanced,1.6300907135009766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,2,balanced,1.7222612698872883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,2,balanced,2.7040799458821616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,2,1,power_law_1.2,0.4052864074707031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,2,balanced,2.9853601455688477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,2,balanced,4.185680071512858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,2,1,power_law_1.2,0.39360001087188723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,2,balanced,5.21343453725179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,2,1,power_law_1.2,0.3999167919158936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,2,1,power_law_1.2,0.4097152233123779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,2,1,power_law_1.2,2.9669952392578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,2,1,power_law_1.2,0.43935360908508303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,2,1,power_law_1.2,0.4772031784057617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,2,1,power_law_1.2,0.5323071956634522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,2,1,power_law_1.2,0.4735551834106445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,2,1,power_law_1.01,0.6493120193481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,2,1,power_law_1.2,0.547980785369873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,2,1,power_law_1.2,0.527558422088623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,2,1,power_law_1.01,0.9831551551818848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,2,1,power_law_1.01,0.83306884765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,2,1,power_law_1.2,0.5787519931793212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,2,1,power_law_1.01,1.160416030883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,2,1,power_law_1.01,0.9902079582214356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,2,1,power_law_1.01,1.4921216011047362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,2,1,power_law_1.01,1.3375743865966796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,2,1,power_law_1.01,1.653945541381836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,2,1,power_law_1.01,1.8520191192626954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,2,1,power_law_1.01,2.3059839248657226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,2,1,power_law_1.01,2.5748672485351562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,2,balanced,6.745546976725261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,2,1,power_law_1.01,2.96878719329834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,2,1,power_law_1.01,3.222502517700195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,2,1,power_law_1.01,4.2612159729003904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,2,1,power_law_1.2,0.6876160144805908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,2,balanced,9.601999918619791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,2,1,power_law_1.01,5.576409530639649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,2,balanced,0.06857066849867503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,2,balanced,0.10150933265686035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,2,balanced,0.16904000441233316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,2,balanced,0.2789600094159444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,2,balanced,0.5092639923095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,2,balanced,0.6325440009435018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,2,balanced,0.7091626326243082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,2,balanced,0.711674690246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,2,balanced,0.6835359732309977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,2,balanced,0.6879146893819174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,2,balanced,0.6610399881998698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,2,balanced,0.670954704284668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,2,1,power_law_1.2,2.041196823120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,2,balanced,0.6750933329264323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,2,balanced,0.6874879995981852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,2,balanced,0.6992159684499105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,2,balanced,0.7118079662322998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,2,balanced,0.7358773549397787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,2,balanced,0.8665706316630045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,2,balanced,0.9030186335245768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,2,balanced,0.9816533724466959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,2,balanced,1.0615359942118328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,2,balanced,1.5970773696899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,2,balanced,0.041850666205088295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,2,balanced,0.05991999804973602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,2,balanced,0.09112000465393066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,2,balanced,1.7780213356018066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,2,balanced,0.15012266238530478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,2,1,power_law_1.2,0.6525184154510498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,2,balanced,0.2640586694081624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,2,balanced,0.32314133644104004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,2,balanced,0.32625067234039307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,2,balanced,0.3327946662902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,2,balanced,2.496405283610026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,2,balanced,0.33640531698862713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,2,balanced,0.3394186496734619
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,2,balanced,0.34519465764363605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,2,1,power_law_1.2,0.7540671825408936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,2,balanced,3.1875734329223633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,2,balanced,0.3522133429845174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,2,balanced,0.34508800506591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,2,balanced,0.3556640148162842
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,2,balanced,0.3646293481190999
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,2,balanced,0.35658133029937744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,2,balanced,0.36869335174560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,2,balanced,0.4082346757253011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,2,balanced,0.39977598190307617
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,2,balanced,0.6951200167338053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,2,balanced,0.5092906554539999
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,2,balanced,1.2203253110249836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,2,balanced,0.8253493309020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,2,balanced,1.1622986793518066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,2,balanced,0.08493333061536153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,2,balanced,0.10794132947921753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,2,balanced,0.18893865744272867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,2,balanced,1.5381813049316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,2,balanced,0.30330665906270343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,2,balanced,0.5373706817626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,2,balanced,0.6582560141881307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,2,balanced,1.739360014597575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,2,power_law_1.2,7.06864013671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,2,balanced,0.6629546483357748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,2,balanced,0.6659573316574097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,2,balanced,1.0454986890157063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,2,balanced,2.7147839864095054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,2,balanced,1.0517333348592122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,2,balanced,1.0607413450876872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,2,power_law_1.01,0.09340159893035889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,2,balanced,1.0725706418355305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,2,power_law_1.01,0.15761280059814453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,2,balanced,1.0825066566467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,2,power_law_1.01,0.20390400886535645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,2,balanced,1.1039679845174153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,2,balanced,1.1209812959035237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,2,balanced,1.1422613461812336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,2,balanced,1.1744373639424641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,2,balanced,1.2366080284118652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,2,balanced,1.2979733149210613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,2,power_law_1.01,0.27635200023651124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,2,power_law_1.01,0.14439680576324462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,2,power_law_1.01,0.3425280094146729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,2,power_law_1.01,0.2360447883605957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,2,balanced,2.1647146542867026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,2,power_law_1.01,0.33345279693603513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,2,power_law_1.01,0.4974720001220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,2,power_law_1.01,0.42268800735473633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,2,balanced,2.279797395070394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,2,power_law_1.01,0.6117631912231445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,2,power_law_1.01,0.5810944080352783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,2,power_law_1.01,0.6171520233154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,2,power_law_1.01,0.5876607894897461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,2,power_law_1.01,0.9110591888427735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,2,balanced,3.2953707377115884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,2,power_law_1.01,0.6207424163818359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,2,power_law_1.01,0.9523967742919922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,2,power_law_1.01,0.6004288196563721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,2,power_law_1.01,1.0163328170776367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,2,power_law_1.01,0.6253312110900879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,2,balanced,4.230426788330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,2,balanced,4.292933464050293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,2,power_law_1.01,0.6420351982116699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,2,power_law_1.01,1.1280896186828613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,2,power_law_1.01,0.6627711772918701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,2,power_law_1.01,1.168678379058838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,2,power_law_1.01,0.6778687953948974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,2,power_law_1.01,1.1199872016906738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,2,power_law_1.01,0.7113855838775635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,2,balanced,5.536133448282878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,2,balanced,5.889290491739909
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,2,power_law_1.01,1.113491153717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,2,power_law_1.01,0.7365119934082032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,2,power_law_1.01,1.1650943756103516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,2,power_law_1.01,0.9066559791564941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,2,power_law_1.01,1.1501567840576172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,2,power_law_1.01,0.9667967796325684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,2,power_law_1.01,1.2322303771972656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,2,power_law_1.01,1.1326080322265626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,2,power_law_1.01,1.233785629272461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,2,power_law_1.01,1.313094425201416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,2,power_law_1.01,1.3284031867980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,2,power_law_1.01,1.6164800643920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,2,power_law_1.01,1.5519231796264648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,2,balanced,7.509759902954102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,2,power_law_1.01,1.9282239913940429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,2,power_law_1.01,1.6834880828857421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,2,1,power_law_1.2,1.0305472373962403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,2,power_law_1.01,1.88799991607666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,2,power_law_1.01,2.163667106628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,2,balanced,10.732245127360025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,2,power_law_1.01,2.680851173400879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,2,power_law_1.01,2.7100223541259765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,2,1,power_law_1.2,1.1921600341796874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,2,power_law_1.01,3.1796991348266603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,2,power_law_1.01,3.3230911254882813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,2,1,power_law_1.2,1.520076847076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,2,power_law_1.01,4.583686447143554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,2,power_law_1.01,4.880012893676758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,2,balanced,13.892890930175781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,2,1,power_law_1.2,1.8732608795166015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,2,power_law_1.01,5.687238311767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,2,power_law_1.01,6.495455932617188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,2,power_law_1.01,0.10158079862594604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,2,power_law_1.2,0.15612159967422484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,2,power_law_1.2,0.20531840324401857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,2,power_law_1.2,0.3231168031692505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,2,power_law_1.2,0.4419072151184082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,2,power_law_1.2,0.5138879776000976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,2,power_law_1.2,0.862668800354004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,2,power_law_1.01,0.18395520448684693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,2,power_law_1.2,0.9642239570617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,2,power_law_1.01,0.2054271936416626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,2,power_law_1.01,0.28297600746154783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,2,power_law_1.2,0.9972288131713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,2,power_law_1.01,0.3804415941238403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,2,power_law_1.2,1.1440383911132812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,2,power_law_1.01,0.5630208015441894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,2,power_law_1.01,0.5689599990844727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,2,power_law_1.2,1.169926357269287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,2,power_law_1.01,0.060652798414230345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,2,power_law_1.01,0.571776008605957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,2,power_law_1.01,0.08559359908103943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,2,power_law_1.2,1.1204352378845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,2,power_law_1.01,0.10347520112991333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,2,power_law_1.01,0.9580479621887207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,2,power_law_1.2,1.139244842529297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,2,power_law_1.01,0.1484671950340271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,2,power_law_1.01,0.9461183547973633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,2,power_law_1.01,0.18165119886398315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,2,power_law_1.2,1.1677568435668946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,2,power_law_1.01,0.2727679967880249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,2,power_law_1.01,1.0035264015197753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,2,power_law_1.01,0.28646399974823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,2,power_law_1.2,1.181612777709961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,2,power_law_1.01,0.9907584190368652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,2,power_law_1.01,0.29947519302368164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,2,power_law_1.01,0.3064768075942993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,2,power_law_1.2,1.188262367248535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,2,power_law_1.01,1.0120512008666993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,2,power_law_1.01,0.30126080513000486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,2,power_law_1.01,0.33556480407714845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,2,power_law_1.2,1.2588543891906738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,2,power_law_1.01,1.0593855857849122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,2,power_law_1.01,0.356825590133667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,2,power_law_1.01,1.1082688331604005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,2,power_law_1.01,0.32988159656524657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,2,power_law_1.2,1.348192024230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,2,power_law_1.01,0.3556096076965332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,2,power_law_1.01,1.1935744285583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,2,power_law_1.01,0.38511359691619873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,2,power_law_1.2,1.6268415451049805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,2,power_law_1.01,0.36065919399261476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,2,power_law_1.01,1.2735360145568848
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,2,power_law_1.01,0.39083518981933596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,2,power_law_1.2,1.717523193359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,2,power_law_1.01,0.41096320152282717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,2,power_law_1.01,1.481011199951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,2,power_law_1.01,0.45717759132385255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,2,power_law_1.2,1.9880767822265626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,2,power_law_1.01,0.6072768211364746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,2,power_law_1.01,1.716659164428711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,2,power_law_1.01,0.6072447776794434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,2,power_law_1.2,2.2675647735595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,2,power_law_1.01,2.1559871673583983
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,2,power_law_1.01,0.7565248012542725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,2,1,power_law_1.2,2.565875244140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,2,power_law_1.01,7.751884460449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,2,power_law_1.01,0.8837375640869141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,2,power_law_1.01,2.580556869506836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,2,power_law_1.01,1.1828479766845703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,2,power_law_1.2,2.6678655624389647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,2,power_law_1.01,1.4816255569458008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,2,power_law_1.01,3.491449737548828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,2,power_law_1.01,2.1490047454833983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,2,power_law_1.2,3.432953643798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,2,power_law_1.01,9.714463806152343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,2,power_law_1.01,4.282233428955078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,2,power_law_1.01,3.074086380004883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,2,power_law_1.2,0.10279680490493774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,2,power_law_1.2,0.14116480350494384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,2,power_law_1.2,0.21194241046905518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,2,power_law_1.2,4.328326416015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,2,power_law_1.2,0.24815359115600585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,2,power_law_1.2,0.2903872013092041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,2,power_law_1.2,0.532371187210083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,2,power_law_1.01,6.238124847412109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,2,power_law_1.2,0.1019711971282959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,2,power_law_1.2,0.5848256111145019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,2,power_law_1.2,0.13317760229110717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,2,power_law_1.2,0.639244794845581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,2,power_law_1.2,0.20360960960388183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,2,power_law_1.2,0.6087103843688965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,2,power_law_1.2,0.28266880512237547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,2,power_law_1.2,5.824051284790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,2,power_law_1.2,0.6207488059997559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,2,power_law_1.2,0.31690239906311035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,2,power_law_1.2,0.6122687816619873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,2,power_law_1.2,0.5421823978424072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,2,power_law_1.2,0.6424767971038818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,2,power_law_1.2,0.5807040214538575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,2,power_law_1.2,0.6451200008392334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,2,power_law_1.2,0.5860864162445069
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,2,1,power_law_1.2,0.9543487548828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,2,power_law_1.01,7.9753471374511715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,2,1,power_law_1.2,2.628620719909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,2,power_law_1.2,0.6556863784790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,2,power_law_1.2,0.9400256156921387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,2,power_law_1.2,0.682700777053833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,2,power_law_1.2,0.9836544036865235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,2,power_law_1.2,0.7066048145294189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,2,power_law_1.2,1.0270527839660644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,2,power_law_1.2,0.7505216121673584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,2,power_law_1.2,1.0534912109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,2,power_law_1.2,0.9060480117797851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,2,power_law_1.2,1.0440447807312012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,2,power_law_1.2,1.118336009979248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,2,power_law_1.2,1.156499195098877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,2,power_law_1.01,11.461203002929688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,2,power_law_1.2,1.2243968009948731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,2,power_law_1.2,1.0052800178527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,2,power_law_1.2,1.3102656364440919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,2,power_law_1.2,1.163315200805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,2,power_law_1.2,1.5359552383422852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,2,power_law_1.2,1.3369791984558106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,2,power_law_1.2,1.7559680938720703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,2,power_law_1.2,1.6628543853759765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,2,power_law_1.2,0.06077439785003662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,2,power_law_1.2,2.200364875793457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,2,power_law_1.2,0.07521920204162598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,2,power_law_1.2,2.0677888870239256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,2,power_law_1.2,0.1081279993057251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,2,power_law_1.2,0.13521280288696289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,2,power_law_1.2,2.7039167404174806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,2,power_law_1.2,0.16935039758682252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,2,power_law_1.01,14.61396484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,2,power_law_1.2,0.26017279624938966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,2,1,balanced,0.08084266881148021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,2,1,balanced,0.14337600270907083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,2,power_law_1.2,0.27266559600830076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,2,1,balanced,0.24331732590993246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,2,1,balanced,0.456986665725708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,2,power_law_1.2,0.2961728096008301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,2,1,balanced,0.8751413027445475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,2,power_law_1.2,0.3077824115753174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,2,power_law_1.2,3.65148811340332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,2,1,balanced,1.0903573036193848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,2,power_law_1.2,2.7214656829833985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,2,power_law_1.2,0.3190848112106323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,2,1,balanced,1.1301120122273762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,2,power_law_1.2,0.3443711996078491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,2,1,balanced,1.1337333520253499
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,2,power_law_1.2,0.36506240367889403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,2,1,balanced,1.252021312713623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,2,power_law_1.2,0.33298559188842775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,2,power_law_1.2,3.488025665283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,2,1,balanced,1.2517279783884685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,2,power_law_1.2,0.36037120819091795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,2,power_law_1.2,4.47749137878418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,2,power_law_1.2,0.39765760898590086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,2,1,balanced,1.228490670522054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,2,power_law_1.2,0.37353599071502686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,2,1,balanced,1.2305920124053955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,2,power_law_1.2,0.4005887985229492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,2,1,balanced,1.2298293113708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,2,power_law_1.2,0.4320511817932129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,2,1,balanced,1.2448000113169353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,2,power_law_1.2,0.46844801902770994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,2,1,balanced,1.2692320346832275
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,2,power_law_1.2,0.5569407939910889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,2,1,balanced,1.2705120245615642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,2,power_law_1.2,6.139648056030273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,2,1,balanced,1.3029332955678303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,2,power_law_1.2,0.6671999931335449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,2,1,power_law_1.2,4.260281753540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,2,1,balanced,1.5245812733968098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,2,power_law_1.2,0.7808896064758301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,2,power_law_1.2,8.542623901367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,2,1,balanced,1.5600214004516602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,2,power_law_1.2,0.9081024169921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,2,1,balanced,1.6117439270019531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,2,1,balanced,1.6396266619364421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,2,1,balanced,2.574735959370931
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,2,power_law_1.2,1.1567551612854003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,2,1,balanced,2.8680480321248374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,2,power_law_1.2,1.4891776084899901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,2,1,balanced,4.195205370585124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,2,1,balanced,5.346709569295247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,2,power_law_1.2,2.1885440826416014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,2,power_law_1.2,8.017017364501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,2,1,balanced,6.789306640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,2,1,balanced,9.041221618652344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,2,1,balanced,0.0740586668252945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,2,1,balanced,0.10269332925478618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,2,1,balanced,0.1742186745007833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,2,1,balanced,0.3012320001920064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,2,1,balanced,0.5480159918467203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,2,1,balanced,0.6323146820068359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,2,1,balanced,0.7910133202870687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,2,1,balanced,0.7354506651560465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,2,1,balanced,0.6779999732971191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,2,1,balanced,0.6796800295511881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,2,1,balanced,0.6501760085423788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,2,1,balanced,0.6554453372955322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,2,1,balanced,0.6548373301823934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,2,1,balanced,0.6643093427022299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,2,1,balanced,0.6725707054138184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,2,1,balanced,0.6810932954152426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,2,1,balanced,0.6949439843495687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,2,1,balanced,0.8341279824574789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,2,1,balanced,0.86189866065979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,2,1,balanced,0.9168426990509033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,2,1,balanced,0.9929707050323486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,2,power_law_1.2,4.646700668334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,2,1,balanced,1.527359962463379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,2,1,balanced,1.6475626627604167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,2,1,balanced,2.3212265968322754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,2,1,balanced,3.0963465372721353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,2,1,balanced,4.015930811564128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,2,power_law_1.2,3.445292663574219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,2,1,balanced,5.53271484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,2,1,balanced,0.07379200061162312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,2,1,balanced,0.10124799609184265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,2,1,balanced,0.17875200510025024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,2,1,balanced,0.30347200234731037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,2,1,balanced,0.5273439884185791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,2,1,balanced,0.652725338935852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,2,1,balanced,0.6547253529230753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,2,1,balanced,0.6577066580454508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,2,1,balanced,1.05567463239034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,2,1,balanced,1.0583946704864502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,2,1,balanced,1.0679093201955159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,2,1,balanced,1.074677308400472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,2,1,balanced,1.0813653469085693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,2,power_law_1.2,11.08843536376953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,2,1,balanced,1.095199982325236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,2,1,balanced,1.107594648996989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,2,1,power_law_1.01,0.08819199800491333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,2,1,balanced,1.1179040273030598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,2,1,power_law_1.01,0.16022399663925171
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,2,1,power_law_1.2,1.3981760025024415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,2,1,power_law_1.01,0.24321279525756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,2,1,balanced,1.1389119625091553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,2,1,power_law_1.01,0.39611520767211916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,2,1,balanced,1.177781343460083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,2,1,balanced,0.043280000487963356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,2,1,power_law_1.01,0.05811840295791626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,2,1,balanced,0.06221333146095276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,2,1,balanced,0.09030399719874065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,2,1,power_law_1.01,0.5334527969360352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,2,1,balanced,0.1486240029335022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,2,1,power_law_1.01,0.08723840117454529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,2,1,balanced,1.218570629755656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,2,1,balanced,0.2656480073928833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,2,1,balanced,0.3253333369890849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,2,1,power_law_1.01,0.858124828338623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,2,1,balanced,0.3297813336054484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,2,1,balanced,0.33644266923268634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,2,1,balanced,2.0879359245300293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,2,1,balanced,0.33710400263468426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,2,1,power_law_1.01,0.9728960037231446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,2,1,balanced,0.33990931510925293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,2,1,balanced,0.3449546496073405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,2,1,balanced,0.35042667388916016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,2,1,balanced,2.1604320208231607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,2,1,power_law_1.01,0.9969599723815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,2,1,balanced,0.39046935240427655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,2,1,balanced,0.4015680154164632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,2,1,balanced,0.4127093156178792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,2,1,power_law_1.01,1.1071807861328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,2,1,balanced,0.41439998149871826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,2,1,balanced,0.4200213352839152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,2,1,balanced,3.138821283976237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,2,1,power_law_1.01,0.15860480070114136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,2,1,balanced,0.4575839837392171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,2,1,power_law_1.01,1.1097472190856934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,2,1,balanced,0.47465598583221436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,2,1,power_law_1.01,0.2301248073577881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,2,1,balanced,0.7393439610799154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,2,1,power_law_1.01,1.0874367713928224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,2,1,power_law_1.01,0.2960576057434082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,2,1,balanced,4.089749336242676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,2,1,balanced,0.5987039804458618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,2,1,power_law_1.01,0.49054718017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,2,1,power_law_1.01,1.098585605621338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,2,1,balanced,1.627679983774821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,2,1,power_law_1.01,0.6026303768157959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,2,1,power_law_1.01,1.1201984405517578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,2,1,balanced,1.113866647084554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,2,1,power_law_1.01,0.6273983955383301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,2,1,power_law_1.01,0.5908095836639404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,2,1,power_law_1.01,1.1372735977172852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,2,1,balanced,1.576624075571696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,2,1,power_law_1.01,0.6110720157623291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,2,1,power_law_1.01,1.1571904182434083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,2,1,balanced,2.0690560340881348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,2,power_law_1.2,5.9212799072265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,2,1,power_law_1.01,0.5879807949066163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,2,1,power_law_1.01,1.1980671882629395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,2,1,power_law_1.01,0.6005311965942383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,2,1,balanced,1.8600799242655437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,2,1,power_law_1.2,5.570188903808594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,2,1,power_law_1.01,0.5957824230194092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,2,1,power_law_1.01,1.251692771911621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,2,power_law_1.2,11.488671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,2,1,power_law_1.01,0.6306367874145508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,2,1,balanced,2.8501758575439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,2,1,balanced,5.188693364461263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,2,1,power_law_1.01,0.6449664115905762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,2,1,power_law_1.01,1.4793024063110352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,2,1,power_law_1.2,0.07916160225868225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,2,1,power_law_1.01,0.6552127838134766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,2,1,power_law_1.01,0.07208319902420043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,2,1,power_law_1.01,1.6051071166992188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,2,1,power_law_1.01,0.0991424024105072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,2,1,power_law_1.01,0.1768447995185852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,2,1,power_law_1.01,0.23541760444641113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,2,1,power_law_1.01,1.818623924255371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,2,1,balanced,7.0653228759765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,2,1,power_law_1.01,0.3263999938964844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,2,1,power_law_1.01,0.5382336139678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,2,1,power_law_1.01,2.0496063232421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,2,1,power_law_1.01,0.5594111919403076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,2,1,power_law_1.2,0.14673279523849486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,2,1,power_law_1.01,0.6909503936767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,2,1,power_law_1.2,0.24021759033203124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,2,1,power_law_1.2,0.3447488069534302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,2,1,power_law_1.01,0.8539903640747071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,2,1,power_law_1.01,2.5508096694946287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,2,1,power_law_1.2,0.48441600799560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,2,1,power_law_1.01,0.9268608093261719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,2,1,power_law_1.2,0.8511167526245117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,2,1,balanced,10.022149403889975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,2,1,power_law_1.01,1.060051155090332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,2,1,power_law_1.01,2.9753536224365233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,2,1,power_law_1.2,0.9601216316223145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,2,1,power_law_1.01,0.04126079976558685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,2,1,power_law_1.01,1.1930368423461915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,2,1,power_law_1.01,0.5651904106140136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,2,1,power_law_1.01,0.059008002281188965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,2,1,power_law_1.2,0.9949055671691894
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,2,1,power_law_1.01,0.08743680119514466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,2,1,power_law_1.01,0.9503040313720703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,2,1,power_law_1.01,1.468876838684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,2,1,power_law_1.2,1.1186559677124024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,2,1,power_law_1.01,0.9484160423278809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,2,1,power_law_1.01,3.968511962890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,2,1,power_law_1.2,1.127999973297119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,2,1,power_law_1.01,1.7399808883666992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,2,1,power_law_1.01,0.9676223754882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,2,1,power_law_1.2,1.138156795501709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,2,1,power_law_1.01,0.9945792198181153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,2,1,power_law_1.2,3.249523162841797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,2,1,power_law_1.01,2.305446434020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,2,1,power_law_1.2,1.1234304428100585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,2,1,power_law_1.01,1.0282367706298827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,2,1,power_law_1.2,0.06238080263137817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,2,1,balanced,12.904319763183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,2,1,power_law_1.2,0.08938239812850952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,2,1,power_law_1.01,0.12617599964141846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,2,1,power_law_1.01,5.0257408142089846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,2,1,power_law_1.2,1.117523193359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,2,1,power_law_1.2,0.15628160238265992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,2,1,power_law_1.01,0.16722559928894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,2,1,power_law_1.2,0.21250560283660888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,2,1,power_law_1.01,0.25938560962677004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,2,1,power_law_1.2,1.1702400207519532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,2,1,power_law_1.01,0.2752511978149414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,2,1,power_law_1.2,0.2824512004852295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,2,1,power_law_1.01,0.29460480213165285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,2,1,power_law_1.2,0.4893184185028076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,2,1,power_law_1.2,1.201529598236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,2,1,power_law_1.01,0.3005120038986206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,2,1,power_law_1.2,0.6050432205200196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,2,1,power_law_1.01,0.29777278900146487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,2,1,power_law_1.2,0.0722815990447998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,2,1,power_law_1.2,1.2267583847045898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,2,1,power_law_1.2,0.6136064052581787
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,2,1,power_law_1.01,0.3245311975479126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,2,1,power_law_1.2,0.09931520223617554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,2,1,power_law_1.01,1.065760040283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,2,1,power_law_1.2,0.5980864048004151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,2,1,power_law_1.2,0.17438080310821533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,2,1,power_law_1.01,6.862214660644531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,2,1,power_law_1.2,1.292262363433838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,2,1,power_law_1.01,3.0179775238037108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,2,1,power_law_1.01,1.142527961730957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,2,1,power_law_1.2,0.22286078929901124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,2,1,power_law_1.2,0.3084160089492798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,2,1,power_law_1.2,1.5570431709289552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,2,1,power_law_1.01,1.1592896461486817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,2,1,power_law_1.2,0.5167871952056885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,2,1,power_law_1.2,0.5530687808990479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,2,1,power_law_1.01,1.2441216468811036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,2,1,power_law_1.2,1.649203109741211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,2,1,power_law_1.01,3.9880321502685545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,2,1,power_law_1.2,0.5645440101623536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,2,1,power_law_1.01,0.3417279958724976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,2,1,power_law_1.01,1.4231871604919433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,2,1,power_law_1.01,0.3673088073730469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,2,1,power_law_1.2,0.6277120113372803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,2,1,power_law_1.01,0.4182591915130615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,2,1,power_law_1.2,0.5911231994628906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,2,1,power_law_1.01,1.6159103393554688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,2,1,power_law_1.01,0.4661568164825439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,2,1,power_law_1.01,8.888448333740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,2,1,power_law_1.2,0.6120960235595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,2,1,power_law_1.01,0.39151999950408933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,2,1,power_law_1.2,0.6164224147796631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,2,1,power_law_1.01,5.377260971069336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,2,1,power_law_1.01,0.44627838134765624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,2,1,power_law_1.01,1.9985727310180663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,2,1,power_law_1.2,0.647603178024292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,2,1,power_law_1.01,0.4514624118804932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,2,balanced,0.09668800234794617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,2,balanced,0.16491199533144632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,2,1,power_law_1.2,0.6675712108612061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,2,1,power_law_1.2,1.874835205078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,2,balanced,0.2540213267008464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,2,balanced,0.43513067563374835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,2,1,power_law_1.01,2.377804756164551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,2,1,power_law_1.2,0.6716351985931397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,2,power_law_1.2,15.168460083007812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,2,balanced,0.8321226437886556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,2,1,power_law_1.2,0.7128640174865722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,2,1,power_law_1.2,2.0522815704345705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,2,balanced,1.607360045115153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,2,balanced,0.06825600067774455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,2,balanced,0.09506666660308838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,2,balanced,0.15065600474675497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,2,balanced,0.26370133956273395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,2,1,power_law_1.01,3.1492223739624023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,2,balanced,2.4671093622843423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,2,balanced,0.4861973524093628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,2,1,power_law_1.2,2.517919921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,2,balanced,0.9276800155639648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,2,1,power_law_1.2,0.9605248451232911
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,2,1,power_law_1.01,0.5160128116607666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,2,balanced,2.6175573666890464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,2,balanced,1.5303680102030437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,2,1,power_law_1.01,0.5412159919738769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,2,1,power_law_1.2,0.9772224426269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,2,balanced,1.534597396850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,2,1,power_law_1.01,0.6277696132659912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,2,1,power_law_1.01,3.905759811401367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,2,balanced,2.7827094395955405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,2,1,power_law_1.2,1.017529582977295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,2,1,power_law_1.2,3.1014848709106446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,2,balanced,1.4508053461710613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,2,1,power_law_1.2,0.8803903579711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,2,1,power_law_1.01,0.7080895900726318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,2,1,power_law_1.2,1.0316736221313476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,2,balanced,1.459760030110677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,2,balanced,2.7715733846028647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,2,1,power_law_1.2,0.9439488410949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,2,1,power_law_1.01,0.8414143562316895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,2,balanced,1.4259413083394368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,2,1,power_law_1.2,1.058784008026123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,2,1,power_law_1.2,1.0807999610900878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,2,balanced,1.4398454030354817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,2,1,power_law_1.2,3.9777793884277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,2,1,power_law_1.2,1.1265600204467774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,2,balanced,1.448794682820638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,2,1,power_law_1.01,5.431833648681641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,2,1,power_law_1.2,1.1574463844299316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,2,balanced,1.460917313893636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,2,balanced,1.471343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,2,1,power_law_1.2,1.1942912101745606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,2,balanced,1.4845706621805828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,2,1,power_law_1.2,5.0157825469970705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,2,1,power_law_1.2,1.2744128227233886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,2,balanced,2.7520319620768228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,2,balanced,1.518330732981364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,2,1,power_law_1.2,1.2157055854797363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,2,1,power_law_1.2,1.4400896072387694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,2,balanced,1.7962826093037922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,2,balanced,2.6640213330586753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,2,1,power_law_1.01,6.901849365234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,2,1,power_law_1.2,1.487014389038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,2,balanced,1.8328746159871419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,2,1,power_law_1.2,1.6641536712646485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,2,balanced,2.7074880599975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,2,balanced,1.9105812708536785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,2,1,power_law_1.01,1.113593578338623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,2,1,power_law_1.2,2.066156768798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,2,balanced,2.756592114766439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,2,balanced,1.9931200345357258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,2,1,power_law_1.2,1.7748735427856446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,2,1,power_law_1.2,0.04137600064277649
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,2,1,power_law_1.2,0.060083198547363284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,2,1,power_law_1.01,1.3997183799743653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,2,balanced,2.155381361643473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,2,balanced,2.8174400329589844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,2,1,power_law_1.2,0.08574079871177673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,2,1,power_law_1.2,2.431987190246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,2,1,power_law_1.2,2.3328128814697267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,2,1,power_law_1.2,0.11682560443878173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,2,balanced,2.3178346951802573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,2,1,power_law_1.01,1.9441408157348632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,2,1,power_law_1.2,0.1595903992652893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,2,balanced,2.829728126525879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,2,1,power_law_1.2,0.26375679969787597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,2,1,power_law_1.01,9.774221038818359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,2,balanced,2.724933306376139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,2,1,power_law_1.2,0.2781759977340698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,2,1,power_law_1.2,3.175904083251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,2,1,power_law_1.2,2.954201507568359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,2,1,power_law_1.2,7.148556518554687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,2,balanced,2.8287200927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,2,1,power_law_1.2,0.29454081058502196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,2,1,power_law_1.01,2.847123146057129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,2,1,power_law_1.2,0.30830719470977785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,2,1,power_law_1.2,0.3117248058319092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,2,balanced,4.001887957255046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,2,balanced,3.3264586130777993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,2,1,power_law_1.2,0.33885440826416013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,2,1,power_law_1.2,3.9262718200683593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,2,1,power_law_1.2,0.3585472106933594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,2,1,power_law_1.2,4.008793640136719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,2,1,power_law_1.2,0.3789439916610718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,2,balanced,3.394181251525879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,2,balanced,4.755541483561198
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,2,1,power_law_1.2,0.4248064041137695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,2,1,power_law_1.2,0.48657917976379395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,2,balanced,3.4400107065836587
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,2,1,power_law_1.2,0.4121088027954102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,2,1,power_law_1.2,0.46496000289916994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,2,1,power_law_1.2,5.32740478515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,2,1,power_law_1.2,0.48119678497314455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,2,balanced,6.361349105834961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,2,1,power_law_1.01,12.61944351196289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,2,1,power_law_1.2,0.5335487842559814
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,2,1,power_law_1.2,0.5578176021575928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,2,power_law_1.01,0.15852799415588378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,2,power_law_1.01,0.08712319731712341
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,2,1,power_law_1.2,0.6423423767089844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,2,power_law_1.01,0.253273606300354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,2,power_law_1.01,0.1492095947265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,2,power_law_1.01,0.4307136058807373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,2,power_law_1.01,0.2618112087249756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,2,balanced,3.576064109802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,2,power_law_1.01,0.5891327857971191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,2,power_law_1.01,0.34508159160614016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,2,1,power_law_1.2,0.7323904037475586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,2,power_law_1.01,0.5540544033050537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,2,balanced,3.5791358947753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,2,1,power_law_1.2,5.356665420532226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,2,power_law_1.01,0.7174079895019532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,2,power_law_1.01,0.870751953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,2,balanced,3.7790187199910483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,2,power_law_1.01,1.0801728248596192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,2,power_law_1.01,0.8553600311279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,2,1,power_law_1.2,0.8673600196838379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,2,power_law_1.01,1.067859172821045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,2,power_law_1.01,1.1723072052001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,2,1,power_law_1.2,1.1438336372375488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,2,power_law_1.01,1.0782400131225587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,2,balanced,0.08067200084527333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,2,power_law_1.01,1.4483263969421387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,2,balanced,0.1063253382841746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,2,balanced,0.1565013329188029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,2,power_law_1.01,1.1391360282897949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,2,balanced,0.274181326230367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,2,balanced,0.5094293355941772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,2,balanced,0.9410026868184408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,2,power_law_1.01,1.1850943565368652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,2,power_law_1.01,1.7064767837524415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,2,balanced,4.329984029134114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,2,balanced,1.3883520762125652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,2,balanced,0.04044266790151596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,2,balanced,0.05857066810131073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,2,power_law_1.01,1.1502719879150392
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,2,balanced,0.08929066856702168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,2,balanced,1.3924585978190105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,2,balanced,0.1442400018374125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,2,balanced,0.2558506727218628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,2,power_law_1.01,2.0714303970336916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,2,balanced,0.4732853174209595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,2,power_law_1.01,1.2214783668518066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,2,balanced,0.6830560366312662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,2,balanced,2.275162696838379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,2,balanced,0.6868693033854166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,2,balanced,6.242560068766276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,2,balanced,0.6904853185017904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,2,power_law_1.01,2.0174144744873046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,2,power_law_1.01,1.271667194366455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,2,balanced,2.28549861907959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,2,1,power_law_1.2,6.807698822021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,2,balanced,0.6939946810404459
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,2,1,power_law_1.2,1.4120448112487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,2,balanced,0.7006879647572836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,2,balanced,0.7092106342315674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,2,balanced,2.2961227099100747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,2,power_law_1.01,1.296390438079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,2,balanced,0.7178186575571696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,2,power_law_1.01,2.1082880020141603
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,2,balanced,0.7319680054982504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,2,balanced,2.3082027435302734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,2,balanced,0.7455413341522217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,2,power_law_1.01,1.3126463890075684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,2,balanced,0.7604533036549886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,2,balanced,7.393322626749675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,2,balanced,0.7504266897837321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,2,power_law_1.01,2.289145660400391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,2,balanced,2.3228960037231445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,2,power_law_1.01,1.5583744049072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,2,balanced,0.7874453067779541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,2,balanced,0.7898613611857096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,2,balanced,2.3440906206766763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,2,balanced,0.8359573682149252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,2,power_law_1.01,2.0968063354492186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,2,power_law_1.01,1.6325439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,2,balanced,0.8619466622670492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,2,balanced,2.3680426279703775
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,2,balanced,0.9229546387990316
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,2,balanced,1.2966986497243245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,2,power_law_1.01,2.2071935653686525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,2,balanced,2.386784076690674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,2,balanced,1.1036746501922607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,2,1,power_law_1.2,1.9630271911621093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,2,balanced,10.139066696166992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,2,balanced,2.4266026814778647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,2,balanced,2.2471946080525718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,2,power_law_1.01,2.312991905212402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,2,balanced,2.4956639607747397
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,2,balanced,1.8441866238911946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,2,power_law_1.01,1.7520128250122071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,2,power_law_1.01,2.399020767211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,2,balanced,2.5593759218851724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,2,balanced,3.047072092692057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,2,power_law_1.01,1.9317440032958983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,2,power_law_1.01,2.3928447723388673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,2,1,power_law_1.2,9.673887634277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,2,power_law_1.01,2.3041664123535157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,2,1,power_law_1.2,2.943724822998047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,2,power_law_1.01,2.7780799865722656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,2,power_law_1.2,0.15601279735565185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,2,power_law_1.01,2.6748672485351563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,2,balanced,2.6797920862833657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,2,power_law_1.2,0.24739201068878175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,2,power_law_1.2,0.37562880516052244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,2,1,power_law_1.2,8.884979248046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,2,power_law_1.01,2.955392074584961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,2,balanced,2.8026132583618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,2,power_law_1.2,0.5299071788787841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,2,power_law_1.01,3.306681442260742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,2,power_law_1.2,0.8626560211181641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,2,balanced,3.0427945454915366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,2,power_law_1.2,0.09009919762611389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,2,power_law_1.01,3.1379776000976562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,2,power_law_1.2,0.1538815975189209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,2,power_law_1.2,1.0445440292358399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,2,power_law_1.2,0.21760640144348145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,2,power_law_1.01,3.807622528076172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,2,power_law_1.2,0.3212287902832031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,2,power_law_1.2,1.3012800216674805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,2,power_law_1.01,3.521004867553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,2,power_law_1.2,0.4915520191192627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,2,power_law_1.2,1.794041633605957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,2,power_law_1.2,0.6082623958587646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,2,power_law_1.01,5.508351898193359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,2,power_law_1.01,3.871775817871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,2,power_law_1.2,0.8203776359558106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,2,power_law_1.2,2.0046207427978517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,2,balanced,4.939584096272786
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,2,power_law_1.01,0.057036799192428586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,2,power_law_1.01,0.10030080080032348
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,2,power_law_1.01,0.08412160277366638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,2,power_law_1.2,1.1279935836791992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,2,power_law_1.01,0.154310405254364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,2,power_law_1.2,2.0526079177856444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,2,power_law_1.01,0.1389248013496399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,2,power_law_1.01,0.27207040786743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,2,power_law_1.2,1.054476833343506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,2,power_law_1.01,4.393958282470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,2,power_law_1.01,0.18424960374832153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,2,balanced,5.398591995239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,2,power_law_1.01,0.36399359703063966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,2,power_law_1.2,1.8699392318725585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,2,power_law_1.01,0.28557438850402833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,2,power_law_1.2,1.082470417022705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,2,power_law_1.01,6.722879791259766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,2,power_law_1.01,0.535968017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,2,power_law_1.01,0.37253119945526125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,2,power_law_1.2,1.078598403930664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,2,power_law_1.2,2.0568767547607423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,2,power_law_1.01,0.4055039882659912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,2,1,power_law_1.2,12.47694091796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,2,power_law_1.2,1.179206371307373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,2,power_law_1.01,5.533951950073242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,2,balanced,7.59657096862793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,2,power_law_1.2,2.1584320068359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,2,power_law_1.2,1.1607295989990234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,2,power_law_1.2,2.192032051086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,2,power_law_1.2,1.2156160354614258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,2,power_law_1.01,0.7108223915100098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,2,power_law_1.01,0.8379839897155762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,2,power_law_1.01,0.5275519847869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,2,power_law_1.2,2.2167232513427733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,2,power_law_1.01,6.587654113769531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,2,power_law_1.01,0.5240064144134522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,2,power_law_1.01,1.0301055908203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,2,balanced,10.22158940633138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,2,power_law_1.01,0.498521614074707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,2,power_law_1.2,2.3510208129882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,2,power_law_1.01,1.7246335983276366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,2,power_law_1.01,0.5670911788940429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,2,power_law_1.2,1.2616191864013673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,2,power_law_1.01,0.562886381149292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,2,power_law_1.2,2.4189823150634764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,2,power_law_1.01,1.7197759628295899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,2,power_law_1.01,0.5929152011871338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,2,power_law_1.2,1.2745280265808105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,2,power_law_1.01,0.6390975952148438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,2,power_law_1.01,8.821068572998048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,2,power_law_1.01,1.9300352096557618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,2,power_law_1.2,1.3412991523742677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,2,power_law_1.2,2.892268753051758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,2,power_law_1.01,0.6760831832885742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,2,power_law_1.01,0.708787202835083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,2,power_law_1.01,1.8713727951049806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,2,balanced,14.544789632161459
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,2,power_law_1.01,0.6795455932617187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,2,power_law_1.2,3.1676799774169924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,2,1,balanced,0.07936533292134602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,2,power_law_1.01,0.7751488208770752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,2,power_law_1.01,1.9119680404663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,2,1,balanced,0.1492959956328074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,2,1,balanced,0.24419732888539633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,2,1,balanced,0.4323626756668091
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,2,power_law_1.01,0.7742976188659668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,2,1,balanced,0.8482293287913004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,2,power_law_1.2,3.484646224975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,2,power_law_1.01,1.9567167282104492
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,2,power_law_1.01,0.8912447929382324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,2,power_law_1.2,1.5812992095947265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,2,power_law_1.01,0.9064319610595704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,2,power_law_1.2,1.679635238647461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,2,power_law_1.01,11.316754913330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,2,power_law_1.2,3.6183040618896483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,2,power_law_1.2,0.10167039632797241
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,2,power_law_1.01,1.0677056312561035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,2,power_law_1.2,1.887775993347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,2,1,balanced,0.06842666864395142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,2,1,balanced,0.09633066256841023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,2,1,balanced,0.14923200011253357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,2,1,balanced,1.6319999694824219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,2,1,balanced,0.2600640058517456
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,2,power_law_1.01,1.1990336418151855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,2,1,balanced,0.4877920150756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,2,1,balanced,0.9405013720194498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,2,power_law_1.2,2.0809919357299806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,2,1,balanced,2.510416030883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,2,power_law_1.2,4.296895980834961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,2,power_law_1.01,1.431443214416504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,2,1,balanced,1.6075305938720703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,2,power_law_1.2,0.055430400371551516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,2,1,balanced,1.6188480059305828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,2,power_law_1.2,2.3913087844848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,2,power_law_1.01,1.7199743270874024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,2,power_law_1.2,0.0837823987007141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,2,1,balanced,1.4919253985087078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,2,power_law_1.2,0.12949119806289672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,2,1,balanced,1.4900533358256023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,2,power_law_1.2,4.7763519287109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,2,power_law_1.2,0.17066240310668945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,2,power_law_1.2,2.670207977294922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,2,1,balanced,1.3718560536702473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,2,power_law_1.01,2.2336191177368163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,2,power_law_1.01,2.057484817504883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,2,power_law_1.2,0.26956160068511964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,2,1,balanced,1.3783200581868489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,2,1,balanced,2.565077304840088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,2,1,balanced,1.3868160247802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,2,power_law_1.2,0.34863359928131105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,2,power_law_1.01,2.0694080352783204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,2,1,balanced,1.3991999626159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,2,power_law_1.2,3.508870315551758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,2,power_law_1.2,0.38098559379577634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,2,1,balanced,2.7387253443400064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,2,power_law_1.01,4.195577621459961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,2,1,balanced,1.4122506777445476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,2,power_law_1.2,5.8254142761230465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,2,power_law_1.2,0.5218175888061524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,2,1,balanced,1.4366559982299805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,2,power_law_1.01,2.164556884765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,2,1,balanced,2.7453972498575845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,2,1,balanced,1.4546292622884114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,2,power_law_1.2,0.5305280208587646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,2,power_law_1.2,4.105868911743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,2,1,balanced,2.697941462198893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,2,1,balanced,1.827290693918864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,2,power_law_1.2,0.5472320079803467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,2,power_law_1.01,2.354431915283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,2,1,balanced,1.8487680753072102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,2,power_law_1.2,0.5504511833190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,2,1,balanced,2.71012274424235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,2,1,balanced,1.9274560610453289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,2,power_law_1.2,0.5483903884887695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,2,power_law_1.2,6.9549507141113285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,2,1,balanced,2.7411626180013022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,2,1,balanced,1.948736031850179
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,2,power_law_1.2,0.5920000076293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,2,power_law_1.2,5.658272171020508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,2,1,balanced,2.7783838907877603
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,2,power_law_1.2,0.6564608097076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,2,1,balanced,2.07423464457194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,2,power_law_1.2,0.7008639812469483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,2,1,balanced,2.225397268931071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,2,1,balanced,2.7875839869181314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,2,1,balanced,0.07389866809050243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,2,1,balanced,0.0965333382288615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,2,1,balanced,0.15026666720708212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,2,power_law_1.2,0.15481599569320678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,2,power_law_1.01,2.5215551376342775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,2,1,balanced,0.2629973292350769
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,2,power_law_1.2,0.7185088157653808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,2,1,balanced,0.4981546799341838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,2,1,balanced,2.50272528330485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,2,power_law_1.2,0.24754559993743896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,2,1,balanced,2.790986696879069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,2,1,balanced,0.9420426686604818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,2,power_law_1.2,0.7132480144500732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,2,power_law_1.2,0.34622719287872317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,2,1,balanced,1.4062560399373372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,2,power_law_1.2,0.8282624244689941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,2,power_law_1.01,2.9138240814208984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,2,1,balanced,2.804938634236654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,2,power_law_1.2,7.190214538574219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,2,power_law_1.2,0.4813568115234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,2,1,balanced,1.4182292620340984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,2,power_law_1.2,9.211763000488281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,2,power_law_1.2,0.806112003326416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,2,1,balanced,2.343935966491699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,2,1,balanced,3.40286922454834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,2,power_law_1.2,0.9232192039489746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,2,power_law_1.01,3.2921409606933594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,2,1,balanced,2.3284640312194824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,2,power_law_1.2,0.9308992385864258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,2,1,balanced,3.458810806274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,2,1,balanced,2.3241705894470215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,2,power_law_1.2,1.146675205230713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,2,1,balanced,2.3302507400512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,2,1,balanced,3.5463733673095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,2,power_law_1.01,4.23372802734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,2,power_law_1.2,1.392972755432129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,2,1,balanced,2.340832074483236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,2,1,balanced,3.600741386413574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,2,power_law_1.2,1.5815936088562013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,2,1,balanced,2.3582773208618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,2,power_law_1.2,11.371123504638671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,2,1,balanced,3.850597381591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,2,1,balanced,2.376373291015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,2,power_law_1.01,4.932038497924805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,2,1,balanced,3.7274560928344727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,2,power_law_1.2,1.880659294128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,2,1,balanced,2.391317367553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,2,1,balanced,4.035696029663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,2,1,balanced,3.8305333455403647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,2,power_law_1.2,0.6525119781494141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,2,1,balanced,2.4182559649149575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,2,power_law_1.2,2.4421247482299804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,2,power_law_1.2,0.7659776210784912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,2,1,balanced,2.46834659576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,2,1,balanced,5.5615997314453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,2,1,power_law_1.01,0.08604159951210022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,2,1,balanced,4.272645314534505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,2,1,power_law_1.01,0.14856319427490233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,2,power_law_1.2,1.0230400085449218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,2,1,balanced,2.5088906288146973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,2,1,power_law_1.01,0.24184958934783934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,2,1,power_law_1.01,0.4296448230743408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,2,1,power_law_1.01,0.05917440056800842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,2,1,power_law_1.01,0.08169599771499633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,2,1,power_law_1.01,0.7765120029449463
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,2,power_law_1.2,4.786540985107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,2,1,power_law_1.01,0.14232959747314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,2,1,balanced,6.299733479817708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,2,1,power_law_1.01,1.0683263778686523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,2,1,balanced,0.039103999733924866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,2,1,balanced,0.05878399809201559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,2,power_law_1.01,6.716307067871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,2,1,balanced,0.08668266733487447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,2,1,balanced,0.1418560047944387
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,2,1,balanced,0.2497119903564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,2,1,balanced,0.4657546679178874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,2,1,power_law_1.01,1.305356788635254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,2,1,balanced,0.688917318979899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,2,1,balanced,0.6871999899546305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,2,1,balanced,0.6914560000101725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,2,1,balanced,2.586736043294271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,2,1,balanced,6.307365417480469
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,2,1,balanced,0.6946933269500732
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,2,1,balanced,0.7004960378011068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,2,1,power_law_1.01,0.24914560317993165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,2,1,balanced,0.705450693766276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,2,1,balanced,2.66593599319458
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,2,1,balanced,0.7143680254618326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,2,1,power_law_1.01,0.42371840476989747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,2,1,balanced,0.7272480328877767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,2,1,balanced,0.7404106458028158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,2,1,balanced,2.8401012420654297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,2,1,power_law_1.01,0.5991104125976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,2,power_law_1.01,8.307027435302734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,2,1,balanced,0.7528586387634277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,2,1,balanced,0.8121653397878011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,2,1,power_law_1.01,1.700819206237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,2,1,balanced,0.8626506328582764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,2,1,balanced,0.8562560081481934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,2,1,balanced,9.45633061726888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,2,1,balanced,0.926688035329183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,2,1,balanced,4.753658612569173
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,2,1,balanced,0.9889919757843018
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,2,1,balanced,1.1124746799468994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,2,1,balanced,1.516650676727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,2,1,balanced,5.076560020446777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,2,1,power_law_1.01,0.8363519668579101
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,2,1,balanced,1.4194080034891765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,2,1,power_law_1.01,1.071897602081299
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,2,1,balanced,3.1052640279134116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,2,power_law_1.01,11.462137603759766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,2,1,power_law_1.01,1.0194111824035645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,2,power_law_1.2,1.7531135559082032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,2,1,balanced,1.978592077891032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,2,1,balanced,7.2140318552653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,2,1,power_law_1.01,1.0622591972351074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,2,power_law_1.2,1.729523277282715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,2,1,balanced,3.2835734685262046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,2,1,power_law_1.01,1.0676351547241212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,2,1,power_law_1.01,1.9172607421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,2,power_law_1.2,1.8491071701049804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,2,1,power_law_1.2,0.07696639895439147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,2,1,power_law_1.01,1.0994303703308106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,2,1,power_law_1.2,0.14670079946517944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,2,1,power_law_1.01,1.9877632141113282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,2,1,power_law_1.2,0.2427072048187256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,2,1,power_law_1.01,1.1015487670898438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,2,power_law_1.2,1.937593650817871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,2,1,balanced,9.635973612467447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,2,1,power_law_1.2,0.42979841232299804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,2,1,power_law_1.01,1.1432448387145997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,2,1,power_law_1.01,1.988652801513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,2,power_law_1.2,1.8312320709228516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,2,1,power_law_1.2,0.681004810333252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,2,1,power_law_1.01,0.0711359977722168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,2,1,power_law_1.01,1.1682751655578614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,2,1,power_law_1.01,2.082950401306152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,2,1,power_law_1.2,0.9700736045837403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,2,1,power_law_1.01,1.191750431060791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,2,power_law_1.01,15.064108276367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,2,1,power_law_1.01,2.1294271469116213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,2,1,power_law_1.01,1.2231488227844238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,2,1,power_law_1.2,1.1905407905578613
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,2,1,power_law_1.01,0.0354559987783432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,2,1,power_law_1.2,0.05596799850463867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,2,1,power_law_1.01,1.5355392456054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,2,1,power_law_1.01,0.05575039982795715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,2,1,power_law_1.01,2.164262390136719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,2,1,power_law_1.2,1.6310464859008789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,2,1,balanced,13.800037384033203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,2,1,power_law_1.01,0.08275840282440186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,2,power_law_1.2,1.8714303970336914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,2,1,power_law_1.01,1.6098751068115233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,2,1,power_law_1.01,0.1374079942703247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,2,1,power_law_1.01,2.1968767166137697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,2,1,power_law_1.2,1.9244287490844727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,2,1,power_law_1.01,0.22156798839569092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,2,power_law_1.2,1.9448192596435547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,2,1,power_law_1.01,1.7437952041625977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,2,1,power_law_1.01,0.3191679954528809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,2,1,power_law_1.01,2.2462976455688475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,2,1,power_law_1.2,1.9343040466308594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,2,1,power_law_1.01,0.38360960483551027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,2,1,power_law_1.2,0.08143360018730164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,2,1,power_law_1.01,1.8594432830810548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,2,1,power_law_1.2,0.1418176054954529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,2,1,power_law_1.01,2.403910446166992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,2,1,power_law_1.2,0.25280640125274656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,2,1,power_law_1.2,1.9869888305664063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,2,1,power_law_1.01,2.1312959671020506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,2,1,power_law_1.01,0.09496960043907166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,2,1,power_law_1.2,0.3904639959335327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,2,1,power_law_1.01,0.14595199823379518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,2,1,power_law_1.2,2.014291191101074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,2,1,power_law_1.01,2.8289472579956056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,2,1,power_law_1.2,0.5412799835205078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,2,1,power_law_1.01,2.3842239379882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,2,1,power_law_1.01,0.2609087944030762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,2,power_law_1.2,2.0325056076049806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,2,1,power_law_1.2,0.07149440050125122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,2,1,power_law_1.01,0.48485121726989744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,2,1,power_law_1.2,0.7321983814239502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,2,1,power_law_1.2,0.0951744019985199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,2,1,power_law_1.2,2.0732160568237306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,2,1,power_law_1.01,0.5062975883483887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,2,1,power_law_1.01,2.857209587097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,2,1,power_law_1.2,0.14714879989624025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,2,power_law_1.2,2.2305856704711915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,2,1,power_law_1.01,2.9851455688476562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,2,1,power_law_1.2,1.0264575958251954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,2,1,power_law_1.01,0.5160831928253173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,2,1,power_law_1.2,0.261299204826355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,2,1,power_law_1.2,2.145849609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,2,1,power_law_1.01,0.5365888118743897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,2,1,power_law_1.2,0.4090240001678467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,2,power_law_1.2,2.3861183166503905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,2,1,power_law_1.01,3.089465522766113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,2,1,power_law_1.01,0.5533887863159179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,2,1,power_law_1.01,3.55797119140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,2,1,power_law_1.2,0.5658112049102784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,2,1,power_law_1.2,2.2405696868896485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,2,1,power_law_1.01,0.5725567817687989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,2,1,power_law_1.2,0.6729536056518555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,2,power_law_1.2,2.6028480529785156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,2,1,power_law_1.01,0.5987775802612305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,2,1,power_law_1.2,2.251078414916992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,2,1,power_law_1.2,0.9284671783447266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,2,1,power_law_1.01,0.6309055805206298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,2,1,power_law_1.2,0.9682880401611328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,2,1,power_law_1.01,4.837273788452149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,2,power_law_1.2,3.0653312683105467
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,2,1,power_law_1.01,0.662227201461792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,2,1,power_law_1.2,1.0453439712524415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,2,1,power_law_1.2,1.5972160339355468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,2,1,power_law_1.2,2.38472957611084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,2,1,power_law_1.01,0.7569471836090088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,2,1,power_law_1.2,1.004742431640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,2,1,power_law_1.01,0.4602560043334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,2,1,power_law_1.01,0.9313216209411621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,2,1,power_law_1.2,1.6885503768920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,2,power_law_1.2,3.412211227416992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,2,1,power_law_1.2,1.066316795349121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,2,1,power_law_1.2,2.8380224227905275
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,2,1,power_law_1.01,0.8179455757141113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,2,1,power_law_1.01,5.968390274047851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,2,1,power_law_1.2,1.723263931274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,2,1,power_law_1.2,1.0954624176025392
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,2,1,power_law_1.01,1.017683219909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,2,1,power_law_1.2,2.929952049255371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,2,1,power_law_1.2,1.1445568084716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,2,1,power_law_1.01,0.9868288040161133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,2,power_law_1.2,4.42334098815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,2,1,power_law_1.01,3.311564636230469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,2,1,power_law_1.2,1.1656767845153808
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,2,1,power_law_1.01,1.2556608200073243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,2,1,power_law_1.2,3.196575927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,2,1,power_law_1.2,1.1936256408691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,2,1,power_law_1.01,1.2412096023559571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,2,1,power_law_1.01,3.709939193725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,2,1,power_law_1.2,1.2483776092529297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,2,power_law_1.2,5.349459075927735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,2,1,power_law_1.2,1.8111040115356445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,2,1,power_law_1.01,1.5804800033569335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,2,1,power_law_1.2,3.4886974334716796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,2,1,power_law_1.2,1.5743231773376465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,2,1,power_law_1.2,1.863225555419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,2,1,power_law_1.01,1.7317312240600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,2,1,power_law_1.01,4.292095947265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,2,1,power_law_1.2,1.645088005065918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,2,1,power_law_1.2,1.9427711486816406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,2,1,power_law_1.2,3.9959232330322267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,2,1,power_law_1.01,0.6439807891845704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,2,1,power_law_1.01,2.2880640029907227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,2,1,power_law_1.2,1.8169599533081056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,2,power_law_1.2,6.775186920166016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,2,1,power_law_1.01,0.7551040172576904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,2,1,power_law_1.2,1.9758783340454102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,2,1,power_law_1.01,5.179193496704102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,2,1,power_law_1.2,2.035615921020508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,2,1,power_law_1.01,0.9630271911621093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,2,1,power_law_1.2,2.011430358886719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,2,1,power_law_1.01,3.487071990966797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,balanced,0.04752000172932943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,balanced,0.06274666885534923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,balanced,0.09801600376764934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,balanced,0.1683680017789205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,2,1,power_law_1.2,2.241856002807617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,balanced,0.2941546638806661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,balanced,0.5465813477834066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,2,1,power_law_1.2,2.163987159729004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,balanced,0.5791039864222208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,balanced,0.5803093512852987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,balanced,0.6186079978942871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,2,1,power_law_1.01,6.099961471557617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,balanced,0.6181386709213257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,2,1,power_law_1.2,2.5503040313720704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,2,power_law_1.2,8.700089263916016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,balanced,0.6100533405939738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,2,1,power_law_1.2,2.349440002441406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,2,1,power_law_1.2,4.576729583740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,balanced,0.6118079821268717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,balanced,0.6146186590194702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,balanced,0.6241493225097656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,2,1,power_law_1.2,0.03687039911746979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,balanced,0.6304213205973307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,2,1,power_law_1.2,0.05745919942855835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,balanced,0.6362186670303345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,2,1,power_law_1.2,3.091315269470215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,balanced,0.6521279811859131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,2,1,power_law_1.2,0.08482559919357299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,balanced,0.7718400160471598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,balanced,0.7897066275278727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,2,1,power_law_1.2,0.1394047975540161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,2,1,power_law_1.2,5.32245101928711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,2,1,power_law_1.2,0.20663039684295653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,2,1,power_law_1.01,8.442131042480469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,2,1,power_law_1.2,3.6589824676513674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,2,1,power_law_1.2,0.29772799015045165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,2,1,power_law_1.2,2.5844608306884767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,2,1,power_law_1.2,0.3579776048660278
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,2,1,power_law_1.2,0.4780735969543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,2,power_law_1.2,11.840863800048828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,balanced,0.8300906817118326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,2,1,power_law_1.2,2.9494272232055665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,2,1,power_law_1.2,6.413471984863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,balanced,0.8706346352895101
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,2,1,power_law_1.2,0.4907072067260742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,2,1,power_law_1.2,4.887948989868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,balanced,0.9552906354268392
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,2,1,power_law_1.2,0.5041408061981201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,balanced,1.0430346330006917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,balanced,1.5810987154642742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,2,1,power_law_1.2,0.5257664203643799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,2,1,power_law_1.2,3.3791423797607423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,balanced,1.7635679244995117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,2,1,power_law_1.01,10.478195190429688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,2,1,power_law_1.2,0.5520768165588379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,balanced,0.04780800143877665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,balanced,0.06605866551399231
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,2,1,power_law_1.2,0.5722559928894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,balanced,0.08865066369374593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,balanced,0.12366933623949687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,2,1,power_law_1.2,6.189023971557617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,balanced,0.19686400890350342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,balanced,0.3169599970181783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,balanced,0.3877013524373372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,2,1,power_law_1.2,4.11596794128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,balanced,0.38916265964508057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,2,1,power_law_1.2,8.458271789550782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,balanced,0.34906665484110516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,balanced,0.3513013521830241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,balanced,0.3516533374786377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,balanced,0.3556266625722249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,balanced,0.359877347946167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,2,1,power_law_1.01,1.6749055862426758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,balanced,0.3699893156687419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,balanced,0.3741813500722249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,balanced,0.38066665331522626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,balanced,0.39432533582051593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,balanced,0.48163731892903644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,2,1,power_law_1.01,1.7087936401367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,2,1,power_law_1.2,4.876748657226562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,balanced,0.49983465671539307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,2,power_law_1.2,15.406431579589844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,2,1,power_law_1.2,0.6101759910583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,balanced,0.5423520008722941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,balanced,0.5843626658121744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,2,1,power_law_1.2,0.6493951797485351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,balanced,0.6708373228708903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,2,1,power_law_1.01,1.8166784286499023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,balanced,0.7587573528289795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,2,1,power_law_1.2,0.6876480102539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,balanced,1.1046613057454426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,2,1,power_law_1.01,1.8554943084716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,balanced,1.2800532976786296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,2,1,power_law_1.2,0.7778624057769775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,2,1,power_law_1.2,10.555129241943359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,balanced,2.452928066253662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,balanced,1.8173227310180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,2,1,power_law_1.2,6.310790252685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,2,1,power_law_1.2,0.9636223793029786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,power_law_1.01,0.06936960220336914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,2,1,power_law_1.01,1.8956863403320312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,balanced,3.1452693939208984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,power_law_1.01,0.10757759809494019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,balanced,2.359498659769694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,power_law_1.01,0.13201279640197755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,2,1,power_law_1.01,1.922092819213867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,power_law_1.01,0.21071999073028563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,power_law_1.01,0.2435136079788208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,balanced,0.029194665451844532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,balanced,0.03143999973932902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,balanced,0.047930667797724404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,2,1,power_law_1.01,1.9711423873901368
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,balanced,0.06937066713968913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,power_law_1.01,0.34104959964752196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,balanced,0.10603732864061992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,balanced,0.17382399241129556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,balanced,0.17704000075658163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,balanced,0.17917333046595255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,power_law_1.01,0.4661248207092285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,balanced,0.18105600277582803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,2,1,power_law_1.2,0.8692352294921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,2,1,power_law_1.2,7.744895935058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,power_law_1.01,0.4399104118347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,2,1,power_law_1.01,2.034579277038574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,power_law_1.01,0.04707199931144714
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,2,1,power_law_1.2,1.0962047576904297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,power_law_1.01,0.5177792072296142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,power_law_1.01,0.5013440132141114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,2,1,power_law_1.01,2.089068794250488
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,2,1,power_law_1.2,1.0320639610290527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,power_law_1.01,0.5058752059936523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,power_law_1.01,0.522054386138916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,2,1,power_law_1.2,1.3119423866271973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,balanced,0.18092266718546549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,2,1,power_law_1.01,2.3008703231811523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,power_law_1.01,0.5247744083404541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,balanced,0.18543465932210287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,balanced,0.18929066260655722
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,balanced,0.1934986710548401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,power_law_1.01,0.5670400142669678
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,balanced,0.19954667488733926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,power_law_1.01,0.06762880086898804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,balanced,0.19430933396021524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,balanced,0.19721599419911703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,power_law_1.01,0.5859519958496093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,balanced,0.20768000682195029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,power_law_1.01,0.09793919920921326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,balanced,0.20960533618927002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,2,1,power_law_1.01,2.4900480270385743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,balanced,0.04716266691684723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,power_law_1.01,0.13093760013580322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,power_law_1.01,0.5694784164428711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,power_law_1.01,0.15869439840316774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,2,1,power_law_1.2,10.503340911865234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,power_law_1.01,0.5935296058654785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,power_law_1.01,0.2120448112487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,power_law_1.01,0.7327104091644288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,2,1,power_law_1.01,2.800576019287109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,power_law_1.01,0.2960832118988037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,power_law_1.01,0.7742335796356201
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,2,1,power_law_1.2,1.3008831977844237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,power_law_1.01,0.05857920050621033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,power_law_1.01,0.29808640480041504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,power_law_1.01,0.2849855899810791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,power_law_1.01,0.8462528228759766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,balanced,0.21987199783325195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,power_law_1.01,0.28621439933776854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,balanced,0.05986666679382324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,balanced,0.07739733159542084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,balanced,0.23296000560124716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,2,1,power_law_1.2,1.642131233215332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,balanced,0.12576533357302347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,power_law_1.01,0.9488320350646973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,balanced,0.18913066387176514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,balanced,0.3390186627705892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,balanced,0.33983465035756427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,balanced,0.34113065401713055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,power_law_1.01,1.1313792228698731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,balanced,0.5561493237813314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,2,1,power_law_1.2,1.799168014526367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,balanced,0.5596426725387573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,balanced,0.5641653140385946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,power_law_1.01,1.2926655769348145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,balanced,0.5716906785964966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,balanced,0.5764586528142294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,balanced,0.5882826646169027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,balanced,0.5989546775817871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,2,1,power_law_1.01,3.1768447875976564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,power_law_1.01,1.6633663177490234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,2,1,power_law_1.2,2.3702911376953124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,balanced,0.2445759971936544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,power_law_1.01,0.3041984081268311
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,balanced,0.3365493218104045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,balanced,0.2967146635055542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,power_law_1.01,0.31984639167785645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,2,1,power_law_1.2,13.291328430175781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,balanced,0.5756906668345133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,power_law_1.01,0.32318079471588135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,power_law_1.01,2.055615997314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,balanced,0.5032106637954712
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,balanced,0.718618631362915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,2,1,power_law_1.01,3.912486267089844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,balanced,0.9311573505401611
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,2,1,power_law_1.2,3.8244991302490234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,power_law_1.2,0.05960959792137146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,power_law_1.01,2.780505561828613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,power_law_1.2,0.04956159889698029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,power_law_1.2,0.0949567973613739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,balanced,0.6075093348821005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,balanced,0.6280959844589233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,power_law_1.2,0.0673088014125824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,power_law_1.2,0.1319808006286621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,power_law_1.01,0.07258880138397217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,balanced,0.6662826538085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,power_law_1.2,0.08728960156440735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,power_law_1.2,0.19005440473556517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,balanced,0.6968479951222738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,power_law_1.01,0.09516159892082214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,power_law_1.2,0.11674239635467529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,2,1,power_law_1.01,4.66640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,power_law_1.2,0.25436160564422605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,power_law_1.01,0.33863680362701415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,power_law_1.01,0.13270399570465088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,power_law_1.01,3.654412841796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,power_law_1.2,0.13134080171585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,power_law_1.01,0.33342719078063965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,power_law_1.2,0.28206720352172854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,power_law_1.01,0.17276159524917603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,power_law_1.2,0.17750400304794312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,power_law_1.01,0.3574399948120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,power_law_1.2,0.4660223960876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,power_law_1.01,0.3667968034744263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,power_law_1.2,0.4451776027679443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,power_law_1.01,0.45415678024291994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,power_law_1.2,0.06032639741897583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,power_law_1.01,0.02971520125865936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,power_law_1.2,0.5345407962799072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,power_law_1.01,0.4833280086517334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,power_law_1.2,0.07189760208129883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,power_law_1.01,0.04611839950084686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,balanced,0.7634666760762533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,power_law_1.2,0.08264319896697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,power_law_1.01,0.5516672134399414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,2,1,power_law_1.01,6.129900741577148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,power_law_1.01,0.055212801694869994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,balanced,0.83188796043396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,power_law_1.2,0.12918399572372435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,power_law_1.01,0.618233585357666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,power_law_1.01,0.07265920042991639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,balanced,1.3198506832122803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,power_law_1.2,0.1659775972366333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,power_law_1.01,0.0890175998210907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,power_law_1.01,0.7625792026519775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,power_law_1.2,0.29241600036621096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,balanced,1.4474666913350422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,power_law_1.2,0.1952191948890686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,power_law_1.01,0.11560319662094116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,power_law_1.2,0.2767359972000122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,power_law_1.2,0.2703360080718994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,power_law_1.01,0.1491711974143982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,power_law_1.2,0.2892159938812256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,power_law_1.01,0.15608960390090942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,power_law_1.2,0.2792383909225464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,power_law_1.2,0.518009614944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,power_law_1.2,0.2834752082824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,power_law_1.01,0.1542400002479553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,power_law_1.2,0.47470722198486326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,power_law_1.2,0.3086143970489502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,power_law_1.2,0.5094848155975342
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,power_law_1.01,0.1541632056236267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,power_law_1.2,0.3082175970077515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,2,1,power_law_1.01,7.637055969238281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,power_law_1.2,0.5429376125335693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,power_law_1.01,0.15791360139846802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,power_law_1.2,0.3302016019821167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,power_law_1.01,0.1666432023048401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,power_law_1.2,0.34250879287719727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,power_law_1.01,0.9138815879821778
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,power_law_1.01,0.16914559602737428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,power_law_1.2,0.35155200958251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,power_law_1.01,0.18558720350265503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,balanced,2.0855840047200522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,power_law_1.01,0.2123136043548584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,power_law_1.2,0.3613248109817505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,power_law_1.01,0.28139519691467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,power_law_1.2,0.38419198989868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,balanced,2.7209653854370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,power_law_1.01,0.2839616060256958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,power_law_1.2,0.47375998497009275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,power_law_1.2,0.48071680068969724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,power_law_1.2,0.5098559856414795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,power_law_1.2,0.47134718894958494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,power_law_1.2,0.5458879947662354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,power_law_1.2,0.5805823802947998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,balanced,3.9760214487711587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,power_law_1.2,0.5776447772979736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,power_law_1.2,0.5739136219024659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,power_law_1.2,0.6490431785583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,power_law_1.01,1.190719985961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,power_law_1.2,0.6015103816986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,power_law_1.2,0.7914944171905518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,power_law_1.2,0.6078720092773438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,power_law_1.01,1.4964032173156738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,balanced,5.2449601491292315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,power_law_1.2,0.9235072135925293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,power_law_1.2,0.7477119922637939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,power_law_1.2,1.222751998901367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,power_law_1.2,0.4978623867034912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,power_law_1.01,1.997542381286621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,power_law_1.2,0.809267234802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,power_law_1.2,0.5055232048034668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,power_law_1.01,0.1775936007499695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,power_law_1.2,0.8856703758239746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,power_law_1.2,1.5245120048522949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,power_law_1.2,0.5277247905731202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,2,1,power_law_1.01,10.519251251220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,power_law_1.01,0.18367359638214112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,power_law_1.01,2.6143680572509767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,power_law_1.2,0.9805503845214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,power_law_1.01,0.1987712025642395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,power_law_1.2,0.5646080017089844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,balanced,0.03789333254098892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,balanced,0.05740800003210703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,balanced,0.0918239951133728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,balanced,0.16170133153597513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,power_law_1.2,0.5751552104949951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,power_law_1.2,2.1051712036132812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,power_law_1.2,1.1694592475891112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,power_law_1.2,0.6236671924591064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,power_law_1.2,1.3718720436096192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,power_law_1.2,0.727513599395752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,power_law_1.2,2.742483139038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,power_law_1.2,0.7964288234710694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,balanced,0.031301334500312805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,balanced,0.042581334710121155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,balanced,0.06599466502666473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,power_law_1.2,1.7656896591186524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,power_law_1.2,0.02922239899635315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,power_law_1.2,0.9672575950622558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,power_law_1.01,0.20867838859558105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,balanced,0.29224000374476117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,power_law_1.2,0.044787201285362246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,power_law_1.01,0.22584319114685059
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,power_law_1.2,0.05400320291519165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,power_law_1.2,2.1793535232543944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,power_law_1.2,1.1239616394042968
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,power_law_1.01,0.24019200801849366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,power_law_1.2,0.07220479846000671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,power_law_1.01,0.2723583936691284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,power_law_1.2,0.08933119773864746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,power_law_1.2,1.4482815742492676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,power_law_1.01,0.32434558868408203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,power_law_1.2,0.11445759534835816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,power_law_1.01,0.386028790473938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,power_law_1.2,2.9468608856201173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,balanced,0.1032373309135437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,2,1,power_law_1.01,13.332806396484376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,balanced,0.1771999994913737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,balanced,0.3137493332227071
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,power_law_1.2,0.1398911952972412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,power_law_1.2,1.803308868408203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,balanced,0.37803200880686444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,balanced,0.3776479959487915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,power_law_1.2,0.14124799966812135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,balanced,0.5521173477172852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,balanced,0.5876959959665934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,power_law_1.2,0.15014400482177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,balanced,0.5881653229395548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,balanced,0.04046933352947235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,power_law_1.01,0.4849088191986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,balanced,0.0517493337392807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,power_law_1.2,0.14854400157928466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,power_law_1.2,3.6394176483154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,balanced,0.07047999898592631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,balanced,0.11989866693814595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,balanced,0.18922134240468344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,power_law_1.2,0.16108160018920897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,power_law_1.01,0.4677887916564941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,balanced,0.3357599973678589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,balanced,0.3345653216044108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,power_law_1.2,0.16656639575958251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,balanced,0.3370026747385661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,power_law_1.01,0.47664642333984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,power_law_1.2,0.1723456025123596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,power_law_1.01,0.49944319725036623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,power_law_1.01,0.4881279945373535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,power_law_1.2,0.18620799779891967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,balanced,0.34994133313496906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,power_law_1.2,0.18108160495758058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,power_law_1.01,0.5380608081817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,balanced,0.3497759898503621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,power_law_1.01,0.6550015926361084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,balanced,0.3413333495457967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,power_law_1.2,0.18586879968643188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,balanced,0.34218132495880127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,power_law_1.01,0.5245823860168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,balanced,0.34257598718007404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,balanced,0.6071093479792277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,power_law_1.2,0.2078336000442505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,power_law_1.2,2.4038463592529298
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,power_law_1.01,0.8478207588195801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,balanced,0.3466720183690389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,balanced,0.6054506699244181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,balanced,0.3518879810969035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,power_law_1.01,0.5319935798645019
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,power_law_1.2,0.2099008083343506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,balanced,0.6010080178578695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,balanced,0.5978933175404867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,power_law_1.2,0.23630719184875487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,power_law_1.01,1.5194879531860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,balanced,0.6001120011011759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,balanced,0.5717013279596964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,balanced,0.6064053376515707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,power_law_1.2,0.2569407939910889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,balanced,0.5722026824951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,power_law_1.01,0.036262398958206175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,balanced,0.6103039979934692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,balanced,0.5765599807103475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,balanced,0.02935466667016347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,power_law_1.2,0.28764159679412843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,balanced,0.033530667424201965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,balanced,0.612175981203715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,balanced,0.5799306631088257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,balanced,0.04671466847260793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,balanced,0.06833066542943318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,balanced,0.6216266552607218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,balanced,0.1032373309135437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,power_law_1.2,0.35240960121154785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,balanced,0.17522666851679483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,balanced,0.7589866320292155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,balanced,0.17669866482416788
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,balanced,0.17974400520324707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,balanced,0.1811359922091166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,power_law_1.2,0.4164544105529785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,balanced,0.18106132745742798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,balanced,0.18481600284576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,balanced,0.3564266761144002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,balanced,0.18735466400782266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,balanced,0.3649919827779134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,balanced,0.19107200702031454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,power_law_1.01,0.5567552089691162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,balanced,0.4649493296941121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,balanced,0.4766346613566081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,balanced,0.5068960189819336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,power_law_1.2,3.049305534362793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,balanced,0.5380959908167521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,power_law_1.01,0.054739201068878175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,balanced,0.5983253320058187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,balanced,0.5806879997253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,balanced,0.6655999819437662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,power_law_1.01,0.09217919707298279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,balanced,0.5905386606852213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,power_law_1.01,0.1598080039024353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,balanced,1.0023679733276367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,balanced,0.7714026769002279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,power_law_1.01,0.2302272081375122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,balanced,1.1305920282999675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,power_law_1.2,0.5289599895477295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,balanced,0.19666133324305216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,power_law_1.2,4.282022476196289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,balanced,1.613034725189209
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,power_law_1.2,0.6531199932098388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,power_law_1.01,0.6131455898284912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,balanced,2.094186623891195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,power_law_1.01,0.6906623840332031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,power_law_1.2,0.8687616348266601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,power_law_1.01,0.030079999566078187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,power_law_1.01,0.7550911903381348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,balanced,0.5970026652018229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,power_law_1.01,0.043609601259231565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,balanced,0.6011733214060465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,balanced,0.7991733551025391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,power_law_1.01,0.06428160071372986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,balanced,0.6144320170084635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,power_law_1.01,0.931123161315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,balanced,0.8308746814727783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,power_law_1.01,0.2942591905593872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,power_law_1.01,0.10117759704589843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,balanced,0.22236265738805136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,balanced,0.8924960295359293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,balanced,0.22613867123921713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,power_law_1.01,1.105241584777832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,power_law_1.01,0.14702719449996948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,balanced,0.2384000023206075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,balanced,0.23894933859507242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,balanced,0.2532800038655599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,power_law_1.01,0.18030719757080077
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,balanced,0.2735626697540283
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,balanced,0.2953866720199585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,power_law_1.01,1.4169983863830566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,power_law_1.01,0.28257920742034914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,balanced,0.4044640064239502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,power_law_1.2,1.5944319725036622
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,balanced,0.37916799386342365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,power_law_1.01,0.28860158920288087
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,balanced,0.8078666528066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,balanced,0.5410186847050985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,power_law_1.01,0.28318080902099607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,balanced,0.6368800004323324
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,balanced,0.7674773534138998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,power_law_1.2,5.672556686401367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,power_law_1.01,0.2892735958099365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,balanced,0.6577173471450806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,balanced,0.9932106335957845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,power_law_1.01,0.43128318786621095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,balanced,0.70087997118632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,power_law_1.01,0.29423999786376953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,power_law_1.01,0.03445119857788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,balanced,0.9581920305887858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,balanced,0.7459626992543539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,power_law_1.2,0.037836799025535585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,power_law_1.01,0.2962048053741455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,power_law_1.01,0.04987519979476929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,power_law_1.01,0.46048641204833984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,power_law_1.2,0.05541759729385376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,balanced,1.2185973326365154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,power_law_1.01,0.30010879039764404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,power_law_1.01,0.06814079880714416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,power_law_1.01,0.4829440116882324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,power_law_1.2,0.09316480159759521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,power_law_1.01,0.31468799114227297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,balanced,1.3095200061798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,power_law_1.01,0.11456639766693115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,power_law_1.2,0.15410560369491577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,power_law_1.01,0.500051212310791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,power_law_1.01,0.3149183988571167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,power_law_1.01,1.7589120864868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,power_law_1.01,0.15548160076141357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,power_law_1.2,0.2099008083343506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,balanced,1.884885311126709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,power_law_1.01,0.33018879890441893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,power_law_1.01,0.19402240514755248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,power_law_1.2,0.28709120750427247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,power_law_1.01,0.3483839988708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,power_law_1.01,0.2532927989959717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,balanced,2.4655307133992515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,power_law_1.01,0.43537278175354005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,power_law_1.2,0.4203455924987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,power_law_1.01,2.4246591567993163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,power_law_1.01,0.2683840036392212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,balanced,1.4491465886433919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,power_law_1.01,0.46598401069641116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,power_law_1.2,0.45128321647644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,power_law_1.01,0.4637184143066406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,power_law_1.01,0.5313280105590821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,balanced,1.5937600135803223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,power_law_1.2,0.4834559917449951
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,power_law_1.01,0.026700800657272337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,balanced,3.5805066426595054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,power_law_1.01,0.4741184234619141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,power_law_1.01,0.580787181854248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,power_law_1.01,0.029772800207138062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,power_law_1.2,0.49219841957092286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,power_law_1.01,0.505625581741333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,power_law_1.01,3.2086849212646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,balanced,2.242431958516439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,power_law_1.01,0.4843455791473389
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,power_law_1.01,0.04692479968070984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,power_law_1.2,0.5044159889221191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,power_law_1.01,0.5141695976257324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,power_law_1.01,0.4909503936767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,power_law_1.01,0.06501759886741638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,power_law_1.2,0.5103424072265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,power_law_1.01,0.5284031867980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,balanced,4.761066754659017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,balanced,2.9765707651774087
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,power_law_1.01,0.08562560081481933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,power_law_1.2,0.5201024055480957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,power_law_1.01,0.5242623805999755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,power_law_1.2,0.027929601073265076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,power_law_1.01,0.11219840049743653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,power_law_1.2,0.5532544136047364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,power_law_1.01,0.5431615829467773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,power_law_1.01,4.338304138183593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,power_law_1.01,0.14005119800567628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,power_law_1.2,0.5621823787689209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,power_law_1.01,0.5590208053588868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,power_law_1.01,0.14765440225601195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,power_law_1.2,0.5621119976043701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,power_law_1.01,0.5698944091796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,power_law_1.01,0.69683837890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,power_law_1.2,0.038438400626182555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,power_law_1.2,0.5950016021728516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,power_law_1.2,0.052934402227401735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,power_law_1.01,0.7092800140380859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,power_law_1.01,0.8153599739074707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,power_law_1.2,0.07128959894180298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,power_law_1.01,0.48791041374206545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,power_law_1.2,0.7424511909484863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,power_law_1.01,0.7475135803222657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,power_law_1.2,0.11307519674301147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,power_law_1.01,1.0590208053588868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,power_law_1.2,0.7773632049560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,power_law_1.01,5.433132934570312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,power_law_1.2,0.04095999896526337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,power_law_1.2,0.13523839712142943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,power_law_1.2,0.0635968029499054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,power_law_1.2,0.1841920018196106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,power_law_1.2,0.8551039695739746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,power_law_1.01,1.2910335540771485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,power_law_1.2,0.09463040232658386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,power_law_1.2,0.24851839542388915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,power_law_1.01,0.14907519817352294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,power_law_1.2,0.12655999660491943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,power_law_1.2,0.9396096229553222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,power_law_1.2,0.2678080081939697
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,power_law_1.01,0.15361920595169068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,power_law_1.2,0.16551680564880372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,power_law_1.01,1.770956802368164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,power_law_1.01,0.1600383996963501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,power_law_1.2,1.0714688301086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,power_law_1.01,0.16520960330963136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,power_law_1.01,0.5184319972991943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,power_law_1.01,0.8143551826477051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,power_law_1.01,0.17268480062484742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,power_law_1.2,1.2414655685424805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,power_law_1.01,2.256991958618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,power_law_1.01,0.5393152236938477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,power_law_1.01,0.18597760200500488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,power_law_1.01,0.8900223731994629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,power_law_1.01,0.554860782623291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,power_law_1.01,0.20120959281921386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,power_law_1.2,1.5673343658447265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,power_law_1.01,0.21867520809173585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,power_law_1.01,1.0499648094177245
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,power_law_1.01,0.2577728033065796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,power_law_1.2,0.02510719895362854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,power_law_1.2,0.4425407886505127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,power_law_1.2,0.26895999908447266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,power_law_1.01,1.19486083984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,power_law_1.01,0.23783040046691895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,power_law_1.2,1.9219648361206054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,power_law_1.2,0.029100799560546876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,power_law_1.2,0.468723201751709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,power_law_1.2,0.2870847940444946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,power_law_1.01,0.28123519420623777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,power_law_1.2,0.046239998936653134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,power_law_1.01,1.5283967971801757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,power_law_1.2,0.27711360454559325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,power_law_1.2,0.4744575977325439
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,power_law_1.01,0.28153600692749026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,power_law_1.2,0.060678398609161376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,power_law_1.2,0.2920639991760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,power_law_1.01,0.3656575918197632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,power_law_1.2,0.07947520017623902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,power_law_1.2,0.4934080123901367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,power_law_1.2,2.590310478210449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,power_law_1.2,0.28436479568481443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,power_law_1.01,0.3499392032623291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,power_law_1.2,0.10459519624710083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,power_law_1.01,0.5793344020843506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,power_law_1.01,1.8540224075317382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,power_law_1.2,0.295411205291748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,power_law_1.01,0.4285696029663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,power_law_1.2,0.1373568058013916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,power_law_1.01,0.6485375881195068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,power_law_1.2,0.3019968032836914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,power_law_1.2,0.1430848002433777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,power_law_1.01,0.5018688201904297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,power_law_1.2,0.3127295970916748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,power_law_1.01,0.7216767787933349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,power_law_1.2,3.2765567779541014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,power_law_1.2,0.14921599626541138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,power_law_1.2,0.3231359958648682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,power_law_1.01,0.6177343845367431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,power_law_1.2,0.1518720030784607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,power_law_1.01,0.8631744384765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,power_law_1.2,0.3378304004669189
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,power_law_1.2,0.1590016007423401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,power_law_1.2,0.3583231925964355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,power_law_1.01,0.9961600303649902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,power_law_1.2,0.16641279458999633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,power_law_1.2,0.5047296047210693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,power_law_1.2,0.45243520736694337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,power_law_1.2,0.17333120107650757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,power_law_1.2,0.5429632186889648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,power_law_1.2,0.1874559998512268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,power_law_1.2,0.488646411895752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,power_law_1.2,0.20246400833129882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,power_law_1.2,0.5455167770385743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,power_law_1.2,0.5430655956268311
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,power_law_1.2,0.22056961059570312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,power_law_1.01,2.5079168319702148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,power_law_1.2,0.562943983078003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,power_law_1.2,0.6052800178527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,power_law_1.2,0.2607872009277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,power_law_1.2,0.60349440574646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,power_law_1.2,0.24675838947296141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,power_law_1.2,0.7217152118682861
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,power_law_1.01,0.8422528266906738
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,power_law_1.2,0.2865216016769409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,power_law_1.2,0.6900800228118896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,power_law_1.2,0.8335424423217773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,power_law_1.2,1.0769408226013184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,power_law_1.2,1.3139007568359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,power_law_1.01,3.1431936264038085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,power_law_1.2,0.28873600959777834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,power_law_1.01,1.279257583618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,power_law_1.2,0.346617603302002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,power_law_1.2,1.8014976501464843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,power_law_1.2,0.7468736171722412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,power_law_1.2,0.3617856025695801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,power_law_1.01,1.5444095611572266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,power_law_1.2,0.43343358039855956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,power_law_1.2,0.8941760063171387
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,power_law_1.2,0.5173632144927979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,power_law_1.2,2.2684864044189452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,power_law_1.2,1.0380672454833983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,power_law_1.01,2.0946176528930662
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,power_law_1.2,0.6299456119537353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,power_law_1.2,1.3128000259399415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,power_law_1.2,0.8624768257141113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,power_law_1.01,1.300115203857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,power_law_1.01,2.6131519317626952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,power_law_1.2,1.566643238067627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,power_law_1.2,1.4102272033691405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,power_law_1.2,2.10199031829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,power_law_1.01,3.6723262786865236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,power_law_1.2,2.6267391204833985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,power_law_1.01,4.689728164672852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,power_law_1.2,3.6581184387207033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,2,balanced,0.039493332306543984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,2,balanced,0.057114665706952415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,2,balanced,0.09128000338872273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,power_law_1.2,4.680255889892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,2,balanced,0.03357866654793421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,2,balanced,0.04398400088151296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,2,balanced,0.06429333488146464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,2,balanced,0.1486026644706726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,2,balanced,0.2384213407834371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,2,balanced,0.3132159908612569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,2,power_law_1.01,0.0531328022480011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,2,power_law_1.01,0.0864960014820099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,2,power_law_1.01,0.03930880129337311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,2,power_law_1.01,0.10373760461807251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,2,power_law_1.01,0.05445759892463684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,2,power_law_1.01,0.0690496027469635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,2,power_law_1.01,0.09830399751663207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,2,balanced,0.10037866234779358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,2,power_law_1.01,0.11543680429458618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,2,balanced,0.14717866977055868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,2,balanced,0.1848106582959493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,2,power_law_1.01,0.1562432050704956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,2,power_law_1.2,0.05364480018615723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,2,power_law_1.01,0.17889280319213868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,2,power_law_1.2,0.03963519930839539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,2,power_law_1.2,0.07505279779434204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,2,power_law_1.2,0.05325440168380737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,2,power_law_1.2,0.11077120304107665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,2,power_law_1.2,0.0700160026550293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,2,balanced,0.33876800537109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,2,balanced,0.3380053440729777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,2,power_law_1.01,0.19397120475769042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,2,power_law_1.01,0.18569600582122803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,2,power_law_1.01,0.14101760387420653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,2,power_law_1.01,0.19139200448989868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,2,power_law_1.01,0.17173759937286376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,2,power_law_1.01,0.19410560131072999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,2,power_law_1.01,0.2570688009262085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,2,power_law_1.01,0.2008960008621216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,2,power_law_1.01,0.20222721099853516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,2,balanced,0.2181546688079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,2,power_law_1.01,0.2111936092376709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,2,power_law_1.01,0.21742720603942872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,2,power_law_1.01,0.22753920555114746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,2,power_law_1.2,0.09343360066413879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,2,power_law_1.01,0.24734079837799072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,2,power_law_1.2,0.13674240112304686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,2,power_law_1.2,0.11867519617080688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,2,power_law_1.01,0.29535999298095705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,2,power_law_1.2,0.18348159790039062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,2,power_law_1.2,0.1496384024620056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,2,power_law_1.01,0.3197439908981323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,2,power_law_1.2,0.25756800174713135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,2,1,balanced,0.03330666571855545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,2,power_law_1.01,0.3831615924835205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,2,1,balanced,0.05553600192070007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,2,1,balanced,0.0848640004793803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,2,power_law_1.01,0.43262720108032227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,2,balanced,0.22021865844726562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,2,balanced,0.364303986231486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,2,power_law_1.01,0.5608767986297607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,2,balanced,0.3680906693140666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,2,power_law_1.01,0.6739071846008301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,2,power_law_1.01,0.9138303756713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,2,power_law_1.01,0.2818624019622803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,2,power_law_1.01,1.1661503791809082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,2,power_law_1.01,0.29745919704437257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,2,1,balanced,0.14776000380516052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,2,power_law_1.2,0.27621119022369384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,2,1,balanced,0.24862933158874512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,2,power_law_1.2,0.18751360177993776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,2,1,balanced,0.3224160075187683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,2,power_law_1.2,0.3102207899093628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,2,power_law_1.2,0.19826560020446776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,2,balanced,0.2125599980354309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,2,balanced,0.21338133017222086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,2,power_law_1.01,0.31322879791259767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,2,balanced,0.3557120164235433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,2,power_law_1.01,0.3205312013626099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,2,1,balanced,0.34585599104563397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,2,1,balanced,0.3482079903284709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,2,power_law_1.2,0.31157760620117186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,2,power_law_1.2,0.3277760028839111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,2,power_law_1.2,0.18547840118408204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,2,power_law_1.2,0.18945280313491822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,2,balanced,0.35645333925882977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,2,balanced,0.35820265611012775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,2,balanced,0.36322665214538574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,2,balanced,0.21233065923055014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,2,balanced,0.36804266770680744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,2,1,balanced,0.029205332199732464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,2,balanced,0.3706773519515991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,2,1,balanced,0.03745600084463755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,2,1,balanced,0.05606399973233541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,2,balanced,0.3803466558456421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,2,1,balanced,0.3553440173467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,2,1,balanced,0.35469333330790204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,2,power_law_1.01,0.32001280784606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,2,power_law_1.01,1.584006404876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,2,power_law_1.2,0.32238719463348386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,2,power_law_1.01,2.0354240417480467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,2,1,balanced,0.09616532921791077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,2,1,balanced,0.1472640037536621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,2,1,balanced,0.18587199846903482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,2,power_law_1.2,0.20188798904418945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,2,balanced,0.21592533588409424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,2,balanced,0.21799467007319132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,2,balanced,0.22466667493184408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,2,balanced,0.23193599780400595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,2,balanced,0.23494933048884073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,2,balanced,0.25013333559036255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,2,balanced,0.4434346755345662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,2,power_law_1.01,0.31880960464477537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,2,balanced,0.45784000555674237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,2,balanced,0.49270931879679364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,2,power_law_1.01,0.321612811088562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,2,balanced,0.5289599895477295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,2,power_law_1.01,0.343507194519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,2,balanced,0.7789333661397299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,2,1,balanced,0.3523999849955241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,2,power_law_1.01,0.34955520629882814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,2,balanced,0.8496426741282145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,2,power_law_1.01,0.3570048093795776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,2,balanced,1.1686186790466309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,2,power_law_1.01,0.3730240106582642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,2,balanced,1.3439572652180989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,2,1,power_law_1.01,0.02991360127925873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,2,1,power_law_1.01,0.0514303982257843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,2,power_law_1.2,0.31171839237213134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,2,power_law_1.2,0.3327615976333618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,2,power_law_1.2,0.34729599952697754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,2,1,balanced,0.2243946592013041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,2,1,balanced,0.22401599089304605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,2,power_law_1.2,0.3635839939117432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,2,power_law_1.2,0.3618240118026733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,2,power_law_1.2,0.39378559589385986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,2,power_law_1.2,0.19867520332336425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,2,power_law_1.2,0.2119231939315796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,2,1,power_law_1.01,0.0816703975200653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,2,power_law_1.2,0.22266240119934083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,2,power_law_1.2,0.22711679935455323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,2,power_law_1.2,0.22961280345916749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,2,balanced,0.292522668838501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,2,balanced,0.3127466638882955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,2,power_law_1.2,0.24849920272827147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,2,balanced,0.34970664978027344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,2,balanced,0.38625065485636395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,2,1,balanced,0.3506293296813965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,2,balanced,0.5591626564661661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,2,1,balanced,0.35315199693044025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,2,1,balanced,0.3551359971364339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,2,balanced,0.631061315536499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,2,1,balanced,0.3590986728668213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,2,1,balanced,0.36090131600697833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,2,1,balanced,0.36769068241119385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,2,power_law_1.01,0.45563521385192873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,2,power_law_1.01,0.4905216217041016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,2,power_law_1.01,0.5534592151641846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,2,balanced,2.0062880516052246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,2,power_law_1.01,0.6203711986541748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,2,balanced,2.507274627685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,2,power_law_1.01,0.7729663848876953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,2,balanced,0.8706239859263102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,2,1,power_law_1.01,0.12243200540542602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,2,1,power_law_1.01,0.02800000011920929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,2,1,power_law_1.01,0.16175999641418456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,2,balanced,1.0242079893747966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,2,1,power_law_1.01,0.03783040046691895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,2,power_law_1.01,0.906009578704834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,2,1,balanced,0.21080533663431802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,2,1,balanced,0.2137920061747233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,2,1,power_law_1.01,0.05559039711952209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,2,1,power_law_1.01,0.0871616005897522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,2,power_law_1.01,1.2172800064086915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,2,1,power_law_1.01,0.10176639556884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,2,power_law_1.2,0.45447678565979005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,2,1,power_law_1.01,0.15105279684066772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,2,power_law_1.2,0.5050559997558594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,2,1,power_law_1.01,0.18933759927749633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,2,power_law_1.01,1.5600192070007324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,2,1,power_law_1.01,0.1869696021080017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,2,power_law_1.2,0.5757440090179443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,2,1,power_law_1.01,0.18692480325698851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,2,power_law_1.2,0.6396351814270019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,2,1,power_law_1.01,0.18966399431228637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,2,1,power_law_1.01,0.1881343960762024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,2,1,power_law_1.01,0.2534847974777222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,2,power_law_1.2,0.8118399620056153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,2,1,power_law_1.01,0.19251199960708618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,2,1,power_law_1.01,0.19173120260238646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,2,power_law_1.2,0.3039999961853027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,2,power_law_1.2,0.9457280158996582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,2,1,power_law_1.01,0.20316801071166993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,2,power_law_1.2,0.332473611831665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,2,1,power_law_1.01,0.20836479663848878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,2,power_law_1.2,0.39200639724731445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,2,1,power_law_1.01,0.21225600242614745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,2,1,power_law_1.01,0.22545280456542968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,2,power_law_1.2,0.44152321815490725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,2,1,power_law_1.01,0.28104960918426514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,2,power_law_1.2,0.5622399806976318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,2,1,power_law_1.01,0.30753920078277586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,2,1,power_law_1.01,0.35024640560150144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,2,power_law_1.2,0.6873663902282715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,2,1,power_law_1.01,0.39715840816497805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,2,1,power_law_1.01,0.4892672061920166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,2,1,balanced,0.43433066209157306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,2,1,balanced,0.4448586702346802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,2,power_law_1.2,1.234553623199463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,2,1,power_law_1.01,0.5849728107452392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,2,1,power_law_1.01,0.7801983833312989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,2,power_law_1.2,0.9428928375244141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,2,1,power_law_1.01,0.9759679794311523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,2,1,power_law_1.01,0.284934401512146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,2,power_law_1.2,1.1455679893493653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,2,1,power_law_1.01,0.29884159564971924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,2,1,balanced,0.2118133306503296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,2,1,power_law_1.01,1.372544002532959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,2,power_law_1.01,2.144191932678223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,2,1,balanced,0.4687360127766927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,2,power_law_1.2,1.5163455963134767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,2,1,balanced,0.4949599901835124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,2,1,power_law_1.01,1.7626304626464844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,2,1,balanced,0.7460052967071533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,2,1,balanced,0.7957546710968018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,2,power_law_1.01,2.7466367721557616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,2,1,balanced,1.1090933481852214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,2,1,balanced,1.2153386274973552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,2,1,balanced,1.8508480389912922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,2,1,balanced,2.309648036956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,2,1,power_law_1.01,0.2927808046340942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,2,balanced,1.540906588236491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,2,1,power_law_1.01,0.29519360065460204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,2,balanced,1.945269266764323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,2,1,power_law_1.2,0.03163520097732544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,2,1,power_law_1.2,0.052172797918319705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,2,1,power_law_1.2,0.027635198831558228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,2,1,power_law_1.2,0.08184319734573364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,2,1,balanced,0.20977065960566202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,2,1,power_law_1.2,0.036800000071525577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,2,1,balanced,0.2113813360532125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,2,1,balanced,0.21423467000325522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,2,1,power_law_1.2,0.05384960174560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,2,1,balanced,0.21762667099634805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,2,1,balanced,0.22197866439819336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,2,1,balanced,0.22819733619689941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,2,balanced,0.15044266978899637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,2,balanced,0.2646560072898865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,2,1,power_law_1.2,0.12169599533081055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,2,1,power_law_1.2,0.1514431953430176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,2,1,power_law_1.2,0.24739840030670165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,2,1,power_law_1.2,0.28273921012878417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,2,1,power_law_1.2,0.2933311939239502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,2,1,power_law_1.2,0.29959681034088137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,2,balanced,0.09474133451779683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,2,1,power_law_1.2,0.2982016086578369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,2,balanced,0.15983466307322183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,2,balanced,0.2683519919713338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,2,1,power_law_1.2,0.3252351999282837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,2,1,power_law_1.2,0.08160640001296997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,2,balanced,0.4625866810480754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,2,1,power_law_1.2,0.31556479930877684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,2,1,power_law_1.2,0.33177599906921384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,2,1,power_law_1.2,0.3423487901687622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,2,1,power_law_1.2,0.34454400539398194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,2,1,power_law_1.01,0.3066368103027344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,2,1,power_law_1.2,0.3478463888168335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,2,1,power_law_1.2,0.3642751932144165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,2,1,power_law_1.2,0.45438079833984374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,2,1,power_law_1.2,0.10033919811248779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,2,1,power_law_1.2,0.4725696086883545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,2,1,balanced,0.2794666687647502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,2,1,power_law_1.2,0.1500159978866577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,2,1,balanced,0.29105067253112793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,2,1,power_law_1.2,0.5311168193817138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,2,power_law_1.2,2.118067169189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,2,balanced,0.49988265832265216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,2,balanced,0.9248480002085367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,2,balanced,0.889077345530192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,2,balanced,1.3273813724517822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,2,balanced,1.6839946111043294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,2,1,balanced,0.31593600908915204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,2,1,balanced,0.3426719903945923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,2,1,power_law_1.2,0.6007872104644776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,2,balanced,2.434256076812744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,2,1,power_law_1.2,0.7276480197906494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,2,1,power_law_1.2,0.8554688453674316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,2,1,power_law_1.01,0.3151808023452759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,2,1,power_law_1.2,1.1204095840454102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,2,1,power_law_1.01,0.31722240447998046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,2,1,power_law_1.2,0.18283519744873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,2,1,power_law_1.01,0.3272255897521973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,2,1,power_law_1.2,0.1935487985610962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,2,1,balanced,0.5145759979883829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,2,1,power_law_1.01,0.3369728088378906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,2,1,balanced,0.5562613407770792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,2,1,power_law_1.01,0.34844160079956055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,2,1,power_law_1.2,1.374828815460205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,2,1,balanced,0.778048038482666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,2,1,power_law_1.01,0.3563008069992065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,2,1,balanced,0.8836320241292318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,2,1,balanced,1.3364267349243164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,2,balanced,1.5212640762329102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,2,1,balanced,1.6661920547485352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,2,balanced,1.5289012591044109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,2,balanced,2.5050880114237466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,2,balanced,2.568277359008789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,2,1,power_law_1.2,0.18778879642486573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,2,1,power_law_1.2,0.19392000436782836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,2,1,power_law_1.01,0.4421247959136963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,2,power_law_1.2,1.704947280883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,2,1,power_law_1.01,0.4691135883331299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,2,1,power_law_1.01,0.5247807979583741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,2,1,power_law_1.2,1.899622344970703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,2,1,power_law_1.01,0.5867136001586915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,2,balanced,1.43285338083903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,2,1,power_law_1.01,0.7002367973327637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,2,balanced,1.441856066385905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,2,1,power_law_1.01,0.8401280403137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,2,balanced,2.7253761291503906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,2,1,power_law_1.01,1.0949567794799804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,2,balanced,2.737253189086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,2,1,power_law_1.01,1.349606418609619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,2,1,power_law_1.01,1.8716224670410155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,2,balanced,0.10212799906730652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,2,1,power_law_1.01,2.4398847579956056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,2,balanced,0.17525333166122437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,2,balanced,0.29205866654713947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,2,1,power_law_1.2,0.19663360118865966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,2,power_law_1.2,2.9281471252441404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,2,balanced,1.4446667035420735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,2,power_law_1.01,0.2547008037567139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,2,power_law_1.01,0.45020160675048826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,2,balanced,0.5196053187052408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,2,power_law_1.01,0.5773119926452637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,2,power_law_1.2,2.1279104232788084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,2,balanced,0.9607733090718588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,2,1,power_law_1.2,2.4248384475708007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,2,balanced,2.70036252339681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,2,power_law_1.01,0.153331196308136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,2,1,power_law_1.2,0.19958399534225463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,2,1,power_law_1.2,0.19788800477981566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,2,power_law_1.01,0.26153600215911865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,2,1,power_law_1.2,0.2093183994293213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,2,balanced,1.4714080492655437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,2,1,power_law_1.2,0.21271040439605712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,2,power_law_1.01,0.3338239908218384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,2,balanced,1.491381327311198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,2,1,power_law_1.2,0.2197376012802124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,2,1,power_law_1.2,0.2297600030899048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,2,balanced,1.521781285603841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,2,power_law_1.01,0.5159872055053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,2,balanced,1.5478240648905437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,2,power_law_1.01,0.6895487785339356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,2,power_law_1.01,0.901420783996582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,2,balanced,1.5937706629435222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,2,power_law_1.01,1.0789376258850099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,2,balanced,1.6495253245035808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,2,power_law_1.01,1.2233856201171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,2,power_law_1.2,0.2552448034286499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,2,balanced,1.3814506530761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,2,power_law_1.2,0.38302719593048096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,2,balanced,1.3947359720865886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,2,power_law_1.01,1.9264192581176758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,2,power_law_1.2,0.5945087909698487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,2,balanced,2.7209173838297525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,2,balanced,1.4028479258219402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,2,power_law_1.2,0.8446847915649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,2,balanced,2.779904047648112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,2,power_law_1.2,1.1393280029296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,2,power_law_1.01,1.3052288055419923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,2,balanced,2.830474535624186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,2,power_law_1.01,1.3332927703857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,2,power_law_1.2,1.8240768432617187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,2,balanced,2.869914690653483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,2,power_law_1.01,1.1999232292175293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,2,1,power_law_1.2,0.2890496015548706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,2,balanced,2.9131787618001304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,2,power_law_1.2,2.0610368728637694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,2,1,power_law_1.2,0.3112704038619995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,2,power_law_1.01,0.166867196559906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,2,power_law_1.01,1.2587776184082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,2,balanced,2.928938547770182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,2,power_law_1.01,0.28626561164855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,2,power_law_1.2,2.0594560623168947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,2,power_law_1.01,1.3061951637268066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,2,balanced,1.9241706530253093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,2,power_law_1.01,2.102732849121094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,2,power_law_1.01,1.347270393371582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,2,power_law_1.2,2.348262405395508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,2,power_law_1.01,1.3474047660827637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,2,1,power_law_1.2,0.3593663930892944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,2,power_law_1.01,2.136569595336914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,2,1,power_law_1.2,0.4096704006195068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,2,power_law_1.2,2.380064010620117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,2,power_law_1.01,1.4012160301208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,2,1,power_law_1.2,0.5017087936401368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,2,power_law_1.01,0.35713920593261717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,2,1,power_law_1.2,0.6040192127227784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,2,power_law_1.01,1.4682559967041016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,2,power_law_1.2,2.4069631576538084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,2,1,power_law_1.2,0.7955520153045654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,2,balanced,2.018357276916504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,2,power_law_1.01,0.5379199981689453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,2,power_law_1.01,1.5032768249511719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,2,balanced,2.2045493125915527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,2,power_law_1.01,0.7041279792785644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,2,power_law_1.2,2.567616081237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,2,power_law_1.01,1.5804479598999024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,2,balanced,2.3983359336853027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,2,power_law_1.01,1.0706624031066894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,2,power_law_1.2,2.5287872314453126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,2,power_law_1.01,1.1267583847045899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,2,balanced,3.615642547607422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,2,balanced,2.3312479654947915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,2,power_law_1.01,1.1783231735229491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,2,power_law_1.2,2.686412811279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,2,balanced,3.3468214670817056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,2,balanced,2.3456053733825684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,2,balanced,4.028160095214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,2,power_law_1.01,2.2916160583496095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,2,power_law_1.01,1.99552001953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,2,balanced,2.3687733014424643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,2,power_law_1.01,1.9252416610717773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,2,power_law_1.2,2.8146175384521483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,2,power_law_1.01,2.097056007385254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,2,balanced,2.398186683654785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,2,balanced,5.770698547363281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,2,power_law_1.01,2.075366401672363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,2,balanced,2.426074663798014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,2,power_law_1.01,2.117728042602539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,2,power_law_1.2,2.865388870239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,2,power_law_1.01,2.3888832092285157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,2,1,power_law_1.2,0.9935168266296387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,2,balanced,3.5418612162272134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,2,power_law_1.01,2.138431930541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,2,balanced,6.595658620198567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,2,1,power_law_1.2,1.3945856094360352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,2,power_law_1.01,2.3942399978637696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,2,power_law_1.2,2.928883171081543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,2,power_law_1.01,2.7334848403930665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,2,balanced,3.6443894704182944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,2,power_law_1.01,2.2067071914672853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,2,balanced,2.4727147420247397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,2,balanced,3.9444907506306968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,2,power_law_1.2,3.569753646850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,2,power_law_1.01,2.2910720825195314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,2,power_law_1.01,3.439321517944336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,2,balanced,2.5149973233540854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,2,balanced,10.149653116861979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,2,power_law_1.01,2.4451967239379884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,2,balanced,2.559253374735514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,2,power_law_1.2,3.900191879272461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,2,balanced,6.003050486246745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,2,power_law_1.01,4.133740615844727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,2,balanced,2.6410239537556968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,2,power_law_1.01,2.5373823165893556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,2,balanced,2.791130701700846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,2,power_law_1.2,4.402905654907227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,2,power_law_1.01,2.7052480697631838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,2,balanced,6.553178787231445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,2,balanced,2.9380534489949546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,2,power_law_1.01,5.646572875976562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,2,balanced,12.663247426350912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,2,power_law_1.01,3.152070426940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,2,power_law_1.2,5.061273574829102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,2,balanced,4.878085454305013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,2,power_law_1.2,0.15475200414657592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,2,power_law_1.01,2.355763244628906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,2,balanced,9.627930959065756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,2,power_law_1.01,3.591603088378906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,2,balanced,5.148885409037272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,2,power_law_1.01,7.166944122314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,2,power_law_1.2,6.266656112670899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,2,power_law_1.01,4.492127990722656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,2,power_law_1.2,0.24011518955230712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,2,balanced,7.462666829427083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,2,power_law_1.2,0.3413439989089966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,2,balanced,11.003823598225912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,2,power_law_1.2,0.48761601448059083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,2,power_law_1.01,5.356774520874024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,2,power_law_1.2,0.6346367835998535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,2,power_law_1.2,7.240013122558594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,2,power_law_1.01,10.140486145019532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,2,power_law_1.2,1.0023743629455566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,2,balanced,8.005669275919596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,2,power_law_1.2,1.2686911582946778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,2,power_law_1.01,2.4803903579711912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,2,power_law_1.2,1.2587136268615722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,2,power_law_1.01,7.296332550048828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,2,power_law_1.01,2.5636928558349608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,2,power_law_1.2,1.2612288475036622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,2,power_law_1.2,1.2406399726867676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,2,power_law_1.2,9.564838409423828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,2,balanced,17.13048044840495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,2,power_law_1.01,2.551180839538574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,2,power_law_1.2,1.3264960289001464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,2,balanced,12.53280512491862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,2,power_law_1.01,2.7065408706665037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,2,power_law_1.01,13.02881317138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,2,1,power_law_1.2,1.7947776794433594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,2,power_law_1.01,8.920972442626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,2,power_law_1.01,2.719014358520508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,2,power_law_1.2,1.3613056182861327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,2,power_law_1.01,2.846329689025879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,2,power_law_1.2,1.3961536407470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,2,power_law_1.2,11.91839370727539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,2,balanced,15.314927419026693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,2,power_law_1.2,1.466329574584961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,2,balanced,21.182384490966797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,2,power_law_1.2,1.5153280258178712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,2,1,balanced,0.09923199812571208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,2,power_law_1.01,12.382495880126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,2,1,balanced,0.15518400073051453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,2,1,balanced,0.26131733258565265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,2,1,balanced,0.48927466074625653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,2,1,balanced,0.9157333374023438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,2,1,balanced,1.3267412980397542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,2,1,balanced,1.6151359875996907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,2,1,balanced,1.644506613413493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,2,power_law_1.2,1.5697728157043458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,2,1,balanced,1.4938507080078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,2,1,balanced,1.5201759338378906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,2,power_law_1.2,1.6767295837402343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,2,1,balanced,1.4296372731526692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,2,power_law_1.2,16.942066955566407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,2,1,balanced,1.461477279663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,2,power_law_1.01,3.292755126953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,2,power_law_1.2,1.9667072296142578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,2,1,balanced,1.4761865933736165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,2,balanced,22.6474609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,2,1,balanced,0.13953600327173868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,2,1,balanced,1.4804320335388184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,2,power_law_1.2,0.16777600049972535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,2,power_law_1.2,2.1457984924316404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,2,1,balanced,1.5001386006673176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,2,power_law_1.01,3.6350399017333985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,2,power_law_1.2,0.23798398971557616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,2,power_law_1.01,16.106573486328124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,2,1,balanced,1.5215199788411458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,2,power_law_1.2,0.3655296087265015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,2,power_law_1.2,2.5395647048950196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,2,1,balanced,1.5636266072591145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,2,1,balanced,1.9509280522664387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,2,power_law_1.01,4.12817268371582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,2,power_law_1.2,2.87007999420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,2,1,balanced,1.9517919222513835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,2,1,balanced,2.0903093020121255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,2,1,balanced,2.2543627421061196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,2,power_law_1.2,3.5822654724121095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,2,1,balanced,3.4737440745035806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,2,power_law_1.2,4.338943862915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,2,1,balanced,3.9171625773111978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,2,power_law_1.01,4.707814407348633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,2,1,balanced,0.24301334222157797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,2,balanced,28.208048502604168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,2,power_law_1.2,0.5036608219146729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,2,1,balanced,5.322725296020508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,2,power_law_1.2,5.787289428710937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,2,power_law_1.2,0.6820032119750976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,2,power_law_1.01,5.832223892211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,2,power_law_1.01,22.751622009277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,2,power_law_1.2,21.363270568847657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,2,1,balanced,6.053946812947591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,2,power_law_1.2,7.455974578857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,2,power_law_1.01,7.068287658691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,2,1,balanced,9.18828264872233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,2,power_law_1.01,9.22279052734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,2,1,power_law_1.01,0.08872320055961609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,2,power_law_1.2,1.098745632171631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,2,1,power_law_1.01,0.1440127968788147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,2,1,balanced,11.359903971354166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,2,power_law_1.2,10.031526184082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,2,power_law_1.2,1.1276288032531738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,2,1,power_law_1.01,0.256166410446167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,2,power_law_1.2,1.1959424018859863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,2,1,balanced,0.4493386745452881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,2,1,power_law_1.01,0.13942400217056275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,2,power_law_1.01,11.886656188964844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,2,power_law_1.01,29.568283081054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,2,1,power_law_1.01,0.4254144191741943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,2,1,power_law_1.01,0.5930240154266357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,2,power_law_1.2,13.143891906738281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,2,1,power_law_1.01,0.9889216423034668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,4,balanced,0.12203733126322429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,4,balanced,0.11487999558448792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,4,balanced,0.17201600472132364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,4,balanced,0.16959466536839804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,4,balanced,0.16819200913111368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,4,balanced,0.16897600889205933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,4,balanced,0.19670933485031128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,4,balanced,0.20068265994389853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,4,balanced,0.23785599072774252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,4,balanced,0.23946134249369302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,4,balanced,0.2539413372675578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,4,balanced,0.25779199600219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,4,balanced,0.25738133986790973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,4,balanced,0.27242134014765423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,2,power_law_1.01,15.784172058105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,2,1,power_law_1.2,0.08912000060081482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,2,1,power_law_1.2,0.14283519983291626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,2,1,power_law_1.2,0.1328703999519348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,2,1,power_law_1.2,0.2564352035522461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,2,1,power_law_1.01,0.2472383975982666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,2,power_law_1.2,2.0253183364868166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,2,1,power_law_1.01,1.2447039604187011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,2,1,power_law_1.01,1.301363182067871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,2,power_law_1.2,2.0281471252441405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,2,1,balanced,0.8567519982655843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,2,power_law_1.2,2.1762624740600587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,2,power_law_1.2,2.2284543991088865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,2,1,power_law_1.2,0.3985984086990356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,2,power_law_1.2,2.242681694030762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,2,1,power_law_1.2,0.5430463790893555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,2,power_law_1.2,2.4090240478515623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,2,power_law_1.01,21.19563446044922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,2,1,power_law_1.2,0.9733887672424316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,4,balanced,0.1021066705385844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,4,balanced,0.0953386624654134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,4,balanced,0.11890133221944173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,4,balanced,0.11967999736467998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,4,balanced,0.1204746663570404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,4,balanced,0.12158399820327759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,4,balanced,0.17003732919692993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,2,power_law_1.2,2.4924863815307616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,4,balanced,0.302511990070343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,4,balanced,0.31333333253860474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,2,1,power_law_1.2,0.24120960235595704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,2,1,power_law_1.01,1.2285568237304687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,4,balanced,0.30848000446955365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,4,balanced,0.37484268347422284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,4,balanced,0.39560532569885254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,4,balanced,0.4864000082015991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,2,1,power_law_1.01,1.2522560119628907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,4,balanced,0.5969546635945638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,4,balanced,0.9115626811981201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,4,balanced,0.1693920095761617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,4,balanced,0.1604426701863607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,4,balanced,1.1309706370035808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,4,balanced,0.16267733772595724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,4,balanced,0.15197867155075073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,4,balanced,0.15686399737993875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,4,balanced,0.16294399897257486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,4,balanced,1.679690678914388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,4,balanced,0.17492266496022543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,2,power_law_1.2,2.619935989379883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,4,balanced,2.2083519299825034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,2,1,power_law_1.2,1.254259204864502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,4,balanced,3.49455992380778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,2,1,power_law_1.01,0.4451648235321045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,2,power_law_1.2,2.8395263671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,2,1,power_law_1.2,1.303206443786621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,4,balanced,4.6952212651570635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,2,1,balanced,1.6363412539164226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,2,power_law_1.2,3.2079872131347655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,2,1,balanced,2.3985279401143393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,2,1,balanced,2.4154720306396484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,2,power_law_1.2,3.7223743438720702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,2,1,balanced,2.454736073811849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,4,balanced,0.12974400321642557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,4,balanced,0.12600533167521158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,4,balanced,0.1452959974606832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,4,balanced,0.14501866698265076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,4,balanced,0.14654399951299033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,2,1,balanced,2.4641599655151367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,4,balanced,0.1709973414738973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,4,balanced,0.15921599666277567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,4,balanced,0.16082132856051126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,4,balanced,0.25143466393152875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,4,balanced,0.2537440061569214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,2,1,power_law_1.01,1.2191935539245606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,4,balanced,0.2590720057487488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,2,power_law_1.2,4.6428478240966795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,2,1,balanced,2.464778741200765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,4,balanced,0.2703946630160014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,4,balanced,0.27451733748118085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,4,balanced,0.29268266757329303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,4,balanced,0.3599840005238851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,4,balanced,0.376693328221639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,4,balanced,0.38704534371693927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,4,balanced,0.48403199513753253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,4,balanced,0.57697065671285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,4,balanced,0.8671147028605143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,2,1,power_law_1.2,1.2274815559387207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,4,balanced,1.0687092940012615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,2,power_law_1.2,5.552371215820313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,2,1,power_law_1.2,0.4484864234924316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,4,balanced,1.560431957244873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,2,1,power_law_1.2,1.2607744216918946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,4,balanced,2.0589119593302407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,4,balanced,3.0632425944010415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,2,power_law_1.2,7.120191955566407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,4,balanced,0.19818667570749918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,4,balanced,3.900021235148112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,4,balanced,0.20704533656438193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,4,balanced,0.20598934094111124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,4,balanced,0.25589332977930707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,4,balanced,0.27931199471155804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,4,balanced,0.36296534538269043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,4,balanced,0.4470986525217692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,2,1,power_law_1.01,1.2375743865966797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,4,balanced,0.6668799718221029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,2,1,power_law_1.01,0.7676095962524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,4,balanced,0.8390293121337891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,2,1,power_law_1.01,1.2618687629699707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,4,balanced,5.891216278076172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,4,balanced,1.2320160071055095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,2,1,power_law_1.01,1.2911999702453614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,4,balanced,1.607269287109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,2,1,power_law_1.01,1.341158390045166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,2,power_law_1.2,9.342591857910156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,4,balanced,2.4197813669840493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,2,1,power_law_1.01,1.3603967666625976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,4,balanced,7.757653554280599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,4,balanced,3.2455358505249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,2,1,power_law_1.01,1.4453887939453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,4,power_law_1.01,0.10929280519485474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,4,power_law_1.01,0.11764479875564575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,4,power_law_1.01,0.1418944001197815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,4,power_law_1.01,0.16111359596252442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,4,power_law_1.01,0.1605504035949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,2,1,power_law_1.2,1.228492832183838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,2,1,balanced,2.46617062886556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,4,power_law_1.01,0.19664000272750853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,4,power_law_1.01,0.1900480031967163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,4,power_law_1.01,0.22099840641021729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,2,1,balanced,2.480122725168864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,4,power_law_1.01,0.27474560737609866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,4,power_law_1.01,0.2857599973678589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,4,power_law_1.01,0.2505408048629761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,2,1,balanced,2.499135971069336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,4,power_law_1.01,0.2813568115234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,4,power_law_1.01,0.2943552017211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,2,power_law_1.2,12.590579223632812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,2,1,balanced,2.533029397328695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,2,1,power_law_1.2,0.7036223888397217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,2,1,balanced,2.556277275085449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,2,1,balanced,2.774282773335775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,4,power_law_1.01,0.3153280019760132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,2,1,power_law_1.01,1.8064767837524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,2,1,power_law_1.2,1.2525952339172364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,2,1,power_law_1.01,1.9627904891967773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,2,1,power_law_1.2,1.3108223915100097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,2,1,power_law_1.01,1.0994815826416016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,2,1,power_law_1.2,1.3514495849609376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,2,1,power_law_1.01,2.261849594116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,2,1,power_law_1.01,1.8009920120239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,2,1,power_law_1.2,1.392416000366211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,2,1,power_law_1.01,2.4849407196044924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,2,1,power_law_1.01,1.9297088623046874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,2,1,power_law_1.2,1.4112064361572265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,2,1,power_law_1.2,1.5161664009094238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,2,power_law_1.2,15.992486572265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,2,1,power_law_1.01,1.9923391342163086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,2,1,power_law_1.01,3.102764892578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,2,1,power_law_1.01,2.052992057800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,2,1,power_law_1.01,3.7802688598632814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,2,1,power_law_1.01,2.0941631317138674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,2,1,power_law_1.2,0.9680447578430176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,2,1,power_law_1.01,4.921574401855469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,2,1,power_law_1.2,1.7704063415527345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,2,1,balanced,2.7348159154256186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,4,balanced,0.04178133110205332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,2,1,power_law_1.2,1.9375104904174805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,2,1,power_law_1.01,6.393945693969727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,2,1,power_law_1.2,1.9802303314208984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,2,1,power_law_1.2,1.8874879837036134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,2,1,power_law_1.2,2.065907287597656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,2,power_law_1.2,22.18023681640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,4,balanced,0.04279999931653341
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,4,balanced,0.05948266883691152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,4,balanced,0.06009600063165029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,2,1,power_law_1.2,2.009984016418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,4,balanced,0.06239999830722809
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,4,balanced,0.062037333846092224
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,4,balanced,0.06492800017197926
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,4,balanced,0.06695466736952464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,2,1,power_law_1.2,2.1225536346435545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,4,balanced,0.09706667065620422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,4,balanced,0.09703466296195984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,4,balanced,0.09081600109736125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,4,balanced,0.23135467370351157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,4,balanced,0.2119093338648478
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,4,balanced,0.18461867173512778
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,4,balanced,0.14452266693115234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,4,balanced,0.109525332848231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,2,1,power_law_1.01,8.987481689453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,4,balanced,0.08591467142105103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,4,balanced,0.10655466715494792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,4,balanced,0.11185066898663838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,4,balanced,0.15054933230082193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,4,balanced,0.17758933703104654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,4,power_law_1.01,0.33843839168548584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,4,balanced,0.24498667319615683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,4,balanced,0.3112906614939372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,4,balanced,0.4452426830927531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,4,balanced,0.5788533290227255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,4,power_law_1.01,0.3565824031829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,4,balanced,0.8509439627329508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,2,1,power_law_1.2,2.2908416748046876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,4,balanced,1.11899201075236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,4,power_law_1.01,0.3915776014328003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,4,power_law_1.01,0.0885312020778656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,4,power_law_1.01,0.099481600522995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,4,power_law_1.01,0.10149760246276855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,4,power_law_1.01,0.11119359731674194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,2,1,power_law_1.2,2.577971267700195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,4,power_law_1.01,0.11292159557342529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,4,power_law_1.01,0.12444159984588624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,4,power_law_1.01,0.4605696201324463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,2,1,power_law_1.01,2.1770303726196287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,4,power_law_1.01,0.15100159645080566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,4,power_law_1.01,0.17543679475784302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,4,power_law_1.01,0.19608319997787477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,2,1,power_law_1.01,11.555506896972656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,2,1,power_law_1.2,3.1533376693725588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,2,1,power_law_1.01,2.246623992919922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,4,power_law_1.01,0.5097663879394532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,4,power_law_1.01,0.1419327974319458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,4,power_law_1.01,0.1363584041595459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,4,power_law_1.01,0.13191039562225343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,2,1,power_law_1.01,2.296998405456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,4,power_law_1.01,0.14391039609909057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,4,power_law_1.01,0.1460927963256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,2,1,power_law_1.2,3.844019317626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,4,power_law_1.01,0.18736000061035157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,4,power_law_1.01,0.19219199419021607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,4,power_law_1.01,0.19611519575119019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,2,1,power_law_1.01,2.4053951263427735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,4,power_law_1.01,0.20209920406341553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,4,power_law_1.01,0.15440640449523926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,4,power_law_1.01,0.722815990447998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,2,1,power_law_1.2,5.039142227172851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,2,1,balanced,2.84117857615153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,4,power_law_1.01,0.16714880466461182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,4,power_law_1.01,0.1841599941253662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,4,power_law_1.01,0.18954240083694457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,2,1,power_law_1.01,2.5390527725219725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,2,1,power_law_1.2,2.203424072265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,2,1,balanced,2.934037208557129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,4,power_law_1.01,0.8884415626525879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,2,power_law_1.2,29.918014526367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,2,1,power_law_1.2,2.316147232055664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,2,1,power_law_1.01,2.668947219848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,2,1,power_law_1.2,6.142419052124024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,2,1,power_law_1.2,2.3509248733520507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,2,1,power_law_1.2,2.5102912902832033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,4,power_law_1.01,0.25589759349823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,4,power_law_1.01,0.2695487976074219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,4,power_law_1.01,0.30076799392700193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,4,power_law_1.01,0.31434240341186526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,2,1,power_law_1.2,2.5871551513671873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,4,power_law_1.01,0.33792641162872317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,4,power_law_1.01,0.3711999893188477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,2,1,power_law_1.2,2.7874944686889647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,4,power_law_1.01,0.41460480690002444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,4,power_law_1.01,0.4584832191467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,4,power_law_1.01,0.6294400215148925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,4,power_law_1.01,1.477184009552002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,2,1,power_law_1.2,8.849100494384766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,4,power_law_1.01,0.7141759872436524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,2,1,power_law_1.01,3.0198080062866213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,2,1,balanced,3.2131945292154946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,4,power_law_1.01,0.9082048416137696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,2,1,balanced,3.429877281188965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,2,1,balanced,5.285999933878581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,4,power_law_1.01,0.2116607904434204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,4,power_law_1.01,1.347334384918213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,4,power_law_1.01,0.21747200489044188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,4,power_law_1.01,0.23992319107055665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,2,1,balanced,5.680655797322591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,4,power_law_1.01,0.2792704105377197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,4,power_law_1.01,0.31706240177154543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,4,power_law_1.01,1.8358335494995117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,2,1,balanced,8.048762639363607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,2,1,power_law_1.01,2.9160320281982424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,4,power_law_1.01,0.5291071891784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,4,power_law_1.01,1.8528255462646483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,2,1,balanced,9.258143742879232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,2,1,power_law_1.01,3.27509765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,2,1,power_law_1.2,3.1563648223876952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,4,power_law_1.01,0.6448959827423095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,4,power_law_1.01,0.04347519874572754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,4,power_law_1.01,2.8164800643920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,4,power_law_1.01,0.04912000000476837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,2,1,power_law_1.2,11.361753845214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,4,power_law_1.01,0.05464320182800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,4,power_law_1.01,0.05761280059814453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,4,power_law_1.2,0.11037440299987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,4,power_law_1.2,0.11769599914550781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,4,power_law_1.2,0.13056639432907105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,4,power_law_1.2,0.16091519594192505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,4,power_law_1.2,0.16033920049667358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,4,power_law_1.2,0.18773119449615477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,2,1,balanced,13.349557240804037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,4,power_law_1.2,0.20259199142456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,4,power_law_1.2,0.2227328062057495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,4,power_law_1.2,0.29111039638519287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,4,power_law_1.2,0.2959872007369995
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,4,power_law_1.01,0.061689597368240354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,4,power_law_1.2,0.2620863914489746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,4,power_law_1.01,0.06307200193405152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,4,power_law_1.2,0.27098879814147947
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,4,power_law_1.01,0.06054400205612183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,4,power_law_1.01,0.06707199811935424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,4,power_law_1.2,0.2998143911361694
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,4,power_law_1.01,0.07576320171356202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,4,power_law_1.01,0.08163840174674988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,4,power_law_1.2,0.3427072048187256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,4,power_law_1.01,0.07783679962158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,4,power_law_1.01,0.1356608033180237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,4,power_law_1.01,0.13771519660949708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,4,power_law_1.01,0.10990719795227051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,4,power_law_1.01,0.13932160139083863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,2,1,balanced,16.525829315185547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,4,power_law_1.01,3.4654399871826174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,4,power_law_1.01,0.11403520107269287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,2,1,power_law_1.01,3.433484649658203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,4,power_law_1.2,0.10614399909973145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,4,power_law_1.01,0.9601152420043946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,4,power_law_1.2,0.11322879791259766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,4,power_law_1.01,0.11570559740066529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,4,power_law_1.2,0.11697280406951904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,4,power_law_1.2,0.1325503945350647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,4,power_law_1.2,0.13281279802322388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,4,power_law_1.2,0.13527040481567382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,4,power_law_1.2,0.15489280223846436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,2,1,power_law_1.01,4.032185745239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,4,power_law_1.2,0.17010560035705566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,4,power_law_1.2,0.19764480590820313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,4,power_law_1.01,0.14776320457458497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,4,power_law_1.2,0.21385600566864013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,4,power_law_1.2,0.16281599998474122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,4,power_law_1.2,0.17952640056610109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,4,power_law_1.2,0.18643200397491455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,4,power_law_1.01,0.17982720136642455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,2,1,power_law_1.01,4.898451232910157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,4,power_law_1.2,0.197817599773407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,4,power_law_1.01,2.9916927337646486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,2,1,power_law_1.01,5.636115264892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,4,power_law_1.01,0.24802560806274415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,2,1,power_law_1.2,3.0473535537719725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,4,power_law_1.01,1.1072256088256835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,2,1,power_law_1.2,3.3468097686767577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,2,1,power_law_1.01,7.534547424316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,4,power_law_1.2,0.33927040100097655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,4,power_law_1.01,0.2969343900680542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,4,power_law_1.2,0.3611840009689331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,4,power_law_1.2,0.3837440013885498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,4,power_law_1.01,0.3774912118911743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,4,power_law_1.01,0.5531263828277588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,4,power_law_1.2,0.4974656105041504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,2,1,power_law_1.01,9.641177368164062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,4,power_law_1.01,1.724825668334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,4,power_law_1.2,0.21631360054016113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,4,power_law_1.2,0.5488063812255859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,4,power_law_1.2,0.24291839599609374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,4,power_law_1.01,6.186086273193359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,2,1,power_law_1.01,13.142649841308593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,4,power_law_1.2,0.24856319427490234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,4,power_law_1.2,0.3002559900283813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,4,power_law_1.2,0.8629504203796386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,4,power_law_1.2,0.389299201965332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,2,1,power_law_1.01,17.122335815429686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,4,power_law_1.01,0.6960256099700928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,2,1,power_law_1.2,3.618617630004883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,4,power_law_1.2,0.1482367992401123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,4,power_law_1.2,0.14908159971237184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,4,power_law_1.2,0.1412287950515747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,4,power_law_1.2,0.1491520047187805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,4,power_law_1.2,0.14997119903564454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,4,power_law_1.01,2.1605247497558593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,4,power_law_1.2,0.19124480485916137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,4,power_law_1.2,0.1892799973487854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,2,1,power_law_1.2,4.032633590698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,4,power_law_1.2,0.1978368043899536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,4,power_law_1.2,0.25664639472961426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,4,power_law_1.2,0.27067520618438723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,4,power_law_1.2,0.5587135791778565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,4,power_law_1.2,0.2907007932662964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,4,power_law_1.2,0.3212287902832031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,4,power_law_1.2,0.32660479545593263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,2,1,power_law_1.2,4.952742385864258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,4,power_law_1.2,0.3526079893112183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,4,power_law_1.2,0.40349440574645995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,4,power_law_1.2,1.1582015991210937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,4,power_law_1.2,0.4630847930908203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,2,1,power_law_1.2,5.9590400695800785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,4,power_law_1.2,0.5638271808624268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,2,1,power_law_1.2,7.868141174316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,4,power_law_1.2,0.7545599937438965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,4,power_law_1.01,4.402067184448242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,4,power_law_1.2,0.9593024253845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,4,power_law_1.2,0.614566421508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,2,1,power_law_1.2,9.92708511352539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,4,power_law_1.01,6.60821762084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,4,power_law_1.2,1.332902431488037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,4,power_law_1.01,3.300825500488281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,4,power_law_1.01,0.9243583679199219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,4,power_law_1.2,0.9128767967224121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,2,1,power_law_1.2,13.393522644042969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,4,power_law_1.2,1.4352383613586426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,4,power_law_1.2,1.7223167419433594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,4,power_law_1.2,2.466534423828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,4,power_law_1.01,1.4993023872375488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,4,power_law_1.01,5.7869312286376955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,2,1,power_law_1.2,16.723635864257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,4,power_law_1.2,1.444979190826416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,4,power_law_1.2,1.9680511474609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,4,power_law_1.01,4.726931381225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,4,1,balanced,0.05097066859404246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,4,1,balanced,0.08536000053087871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,4,1,balanced,0.14417599638303122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,4,1,balanced,0.1405173341433207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,4,1,balanced,0.14223466316858926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,4,1,balanced,0.14460800091425577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,4,1,balanced,0.14626666903495789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,4,1,balanced,0.14458133776982626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,4,1,balanced,0.15128533045450845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,4,1,balanced,0.154639999071757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,4,1,balanced,0.17092265685399374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,4,1,balanced,0.17057067155838013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,4,1,balanced,0.17314134041468301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,4,1,balanced,0.17883199453353882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,4,1,balanced,0.22562134265899658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,4,1,balanced,0.2323626677195231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,4,1,balanced,0.22467732429504395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,4,power_law_1.2,1.6506944656372071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,4,power_law_1.2,3.6012222290039064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,4,power_law_1.2,0.04596480131149292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,4,power_law_1.01,10.417049407958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,4,power_law_1.2,0.05045120120048523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,4,power_law_1.2,0.055743998289108275
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,4,power_law_1.2,0.05868160128593445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,4,power_law_1.2,0.06414719820022582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,4,power_law_1.2,0.06556159853935242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,4,power_law_1.2,0.06546559929847717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,4,power_law_1.2,0.06857600212097167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,4,power_law_1.2,0.08700799942016602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,4,power_law_1.2,0.08209919929504395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,4,1,balanced,0.27747732400894165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,4,power_law_1.2,0.077702397108078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,4,1,balanced,0.28643200794855755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,4,1,balanced,0.377461314201355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,4,power_law_1.2,0.16956160068511963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,4,1,balanced,0.47912001609802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,4,1,balanced,0.6875519752502441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,4,power_law_1.2,0.1258944034576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,4,1,balanced,0.8473920027414957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,4,power_law_1.2,0.12758400440216064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,4,1,balanced,1.2565173308054607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,4,1,balanced,1.603376070658366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,4,power_law_1.2,0.1574336051940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,4,1,balanced,2.4937334060668945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,4,power_law_1.2,0.17523839473724365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,4,1,balanced,3.5257174173990884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,4,power_law_1.2,0.12338559627532959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,4,1,balanced,0.04645866652329763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,4,1,balanced,0.07037866612275441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,4,1,balanced,0.09706667065620422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,4,power_law_1.01,2.0780864715576173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,4,1,balanced,0.09662399689356486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,4,1,balanced,0.09776533643404643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,4,power_law_1.2,5.024172973632813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,4,1,balanced,0.10397332906723022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,4,1,balanced,0.05817066629727682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,4,1,balanced,0.0703306645154953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,4,1,balanced,0.09525332848230998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,4,power_law_1.2,3.0183168411254884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,4,power_law_1.2,0.14632960557937622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,4,1,balanced,0.13356266419092813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,4,1,balanced,0.11343466242154439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,4,1,balanced,0.10724266370137532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,4,1,balanced,0.11589333415031433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,4,1,balanced,0.10700800021489461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,4,1,balanced,0.1086133321126302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,4,1,balanced,0.10812266667683919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,4,1,balanced,0.11512533823649089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,4,1,balanced,0.14435733358065286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,4,1,balanced,0.14789332946141562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,4,power_law_1.2,0.16841599941253663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,4,1,balanced,0.14174933234850565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,4,1,balanced,0.17349867026011148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,4,1,balanced,0.1813546617825826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,4,1,balanced,0.23684799671173096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,4,1,balanced,0.29844266176223755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,4,1,balanced,0.41838399569193524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,4,1,balanced,0.5201919873555502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,4,1,balanced,0.7595413525899252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,4,1,balanced,0.9706453482309977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,4,1,balanced,1.4491359392801921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,4,1,balanced,0.09426132837931316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,4,power_law_1.01,10.352518463134766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,4,1,balanced,0.09502933422724406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,4,power_law_1.2,7.087635040283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,4,1,balanced,1.900544007619222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,4,1,power_law_1.01,0.049779200553894044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,4,1,power_law_1.01,0.06826239824295044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,4,1,power_law_1.01,0.09019520282745361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,4,1,power_law_1.01,0.14122240543365477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,4,1,power_law_1.01,0.13874560594558716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,4,1,power_law_1.01,0.14611200094223023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,4,1,power_law_1.01,0.1426367998123169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,4,1,power_law_1.01,0.14513920545578002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,4,1,power_law_1.01,0.15745279788970948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,4,1,power_law_1.01,0.17267199754714965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,4,1,power_law_1.01,0.16750080585479737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,4,1,power_law_1.01,0.16561919450759888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,4,1,power_law_1.01,0.1698815941810608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,4,1,power_law_1.01,0.17651840448379516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,4,1,power_law_1.01,0.2039167881011963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,4,power_law_1.2,0.24822399616241456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,4,1,power_law_1.01,0.232857608795166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,4,1,power_law_1.01,0.2388159990310669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,4,1,balanced,0.02739733209212621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,4,1,balanced,0.03743999948104223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,4,1,balanced,0.057162667314211525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,4,1,balanced,0.09968533118565877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,4,1,balanced,0.05771199862162272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,4,1,balanced,0.10099732875823975
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,4,1,balanced,0.056746666630109154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,4,1,balanced,0.0993226667245229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,4,1,balanced,0.06011199951171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,4,1,balanced,0.0642080008983612
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,4,1,balanced,0.06612800061702728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,4,1,balanced,0.0683786670366923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,4,1,balanced,0.06909333169460297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,4,1,balanced,0.07005333403746287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,4,1,balanced,0.07230400045712788
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,4,1,balanced,0.07249600191911061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,4,1,balanced,0.0746559997399648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,4,1,balanced,0.11860799789428711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,4,1,balanced,0.10287466645240784
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,4,1,balanced,0.08683199683825175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,4,1,balanced,0.18623999754587808
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,4,1,balanced,0.11065600315729777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,4,1,balanced,0.1401653289794922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,4,1,balanced,0.16963199774424234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,4,1,balanced,0.2312320073445638
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,4,1,balanced,0.2935733397801717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,4,1,balanced,0.42020801703135174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,4,1,balanced,0.5508213440577189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,4,1,power_law_1.01,0.28168959617614747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,4,1,balanced,0.806325356165568
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,4,1,balanced,1.0592052936553955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,4,1,power_law_1.01,0.04048640131950378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,4,1,power_law_1.01,0.05241600275039673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,4,1,power_law_1.01,0.31781759262084963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,4,1,power_law_1.01,0.06048640012741089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,4,1,power_law_1.01,0.082259202003479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,4,1,power_law_1.01,0.08356480002403259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,4,1,power_law_1.01,0.08804479837417603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,4,1,power_law_1.01,0.10748159885406494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,4,1,power_law_1.01,0.11169279813766479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,4,1,power_law_1.01,0.11716480255126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,4,1,power_law_1.01,0.12117760181427002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,4,power_law_1.2,0.2649856090545654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,4,1,power_law_1.01,0.4106175899505615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,4,1,power_law_1.01,0.10370559692382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,4,1,power_law_1.01,0.1053056001663208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,4,1,power_law_1.01,0.1089087963104248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,4,1,power_law_1.01,0.11660799980163575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,4,1,power_law_1.01,0.13178240060806273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,4,1,power_law_1.01,0.1486143946647644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,4,1,power_law_1.01,0.5156544208526611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,4,1,power_law_1.01,0.14746880531311035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,4,1,power_law_1.01,0.17412480115890502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,4,1,power_law_1.01,0.7050367832183838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,4,power_law_1.01,12.59332504272461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,4,1,balanced,0.144378662109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,4,1,power_law_1.01,0.2006592035293579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,4,1,balanced,0.14658666650454202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,4,1,power_law_1.01,0.05901439785957337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,4,1,balanced,0.14892266194025675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,4,1,balanced,0.15291733543078104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,4,1,power_law_1.01,0.06753280162811279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,4,1,balanced,0.15624533096949259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,4,1,balanced,0.163674662510554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,4,1,power_law_1.01,0.07613440155982971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,4,1,balanced,0.25570134321848553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,4,1,balanced,0.2677759925524394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,4,1,power_law_1.01,0.08967679738998413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,4,1,balanced,0.2642186681429545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,4,1,power_law_1.01,0.0915008008480072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,4,1,balanced,0.3438933293024699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,4,1,power_law_1.01,0.09455360174179077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,4,1,balanced,0.4478986660639445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,4,1,power_law_1.01,0.10417280197143555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,4,1,balanced,0.6276959975560507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,4,1,power_law_1.01,0.11711360216140747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,4,1,balanced,0.7876959641774496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,4,1,power_law_1.01,0.1434175968170166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,4,1,balanced,1.1472319761912029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,4,1,balanced,1.4690720240275066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,4,1,balanced,2.167792002360026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,4,1,power_law_1.01,0.9083647727966309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,4,1,balanced,2.792634646097819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,4,1,power_law_1.01,0.26911358833312987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,4,1,power_law_1.01,0.14487680196762084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,4,1,power_law_1.01,0.15900800228118897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,4,1,power_law_1.01,0.20232961177825928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,4,1,power_law_1.01,0.2056447982788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,4,1,balanced,4.143829345703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,4,1,power_law_1.01,0.21960320472717285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,4,power_law_1.2,0.3965696096420288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,4,1,power_law_1.01,0.2380160093307495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,4,1,power_law_1.01,0.31741440296173096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,4,1,power_law_1.01,0.2582079887390137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,4,1,balanced,5.408058802286784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,4,1,power_law_1.01,0.3001919984817505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,4,1,power_law_1.01,0.39153280258178713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,4,1,power_law_1.01,0.46390399932861326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,4,power_law_1.2,3.7773056030273438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,4,1,power_law_1.01,0.6433152198791504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,4,1,power_law_1.01,1.2871487617492676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,4,1,power_law_1.01,0.8187647819519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,4,1,power_law_1.01,0.02961280047893524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,4,1,power_law_1.01,0.035206401348114015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,4,1,power_law_1.01,0.038412800431251524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,4,1,power_law_1.01,0.05424000024795532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,4,1,power_law_1.01,0.05480319857597351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,4,power_law_1.2,2.5328128814697264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,4,1,power_law_1.01,0.05862399935722351
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,4,1,power_law_1.01,0.06318719983100891
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,4,1,power_law_1.01,0.06824319958686828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,4,power_law_1.2,0.5339712142944336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,4,1,power_law_1.01,0.06425600051879883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,4,1,power_law_1.01,0.4329279899597168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,4,1,power_law_1.01,0.06645119786262513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,4,1,power_law_1.01,0.0729088008403778
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,4,1,power_law_1.01,0.06936320066452026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,4,1,power_law_1.01,0.0723904013633728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,4,1,power_law_1.01,0.0815168023109436
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,4,1,power_law_1.01,0.07966079711914062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,4,1,power_law_1.01,0.08481280207633972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,4,1,power_law_1.01,0.09178879857063293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,4,1,power_law_1.01,0.11964800357818603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,4,1,power_law_1.01,1.1490559577941895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,4,1,power_law_1.01,0.1377408027648926
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,4,1,power_law_1.01,0.15712640285491944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,4,1,power_law_1.01,0.17849600315093994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,4,power_law_1.2,9.321485137939453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,4,1,power_law_1.01,1.6785791397094727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,4,1,power_law_1.01,0.2519040107727051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,4,1,power_law_1.01,0.5484032154083252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,4,1,power_law_1.01,1.4822400093078614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,4,1,power_law_1.01,2.1428735733032225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,4,1,power_law_1.01,0.31189761161804197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,4,power_law_1.2,5.733606338500977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,4,power_law_1.2,0.781715202331543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,4,1,power_law_1.01,0.4414912223815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,4,1,power_law_1.01,2.8181951522827147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,4,1,power_law_1.01,2.459347152709961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,4,1,power_law_1.01,0.7785344123840332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,4,power_law_1.2,3.63757438659668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,4,1,power_law_1.01,4.127462387084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,4,1,power_law_1.01,0.5639808177947998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,4,1,power_law_1.01,3.251961517333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,4,1,power_law_1.2,0.05057920217514038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,4,1,power_law_1.2,0.07304959893226623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,4,1,power_law_1.2,0.08859519958496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,4,1,power_law_1.2,0.14319360256195068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,4,1,power_law_1.2,0.14357759952545165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,4,1,power_law_1.2,0.14924800395965576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,4,1,power_law_1.2,0.14360320568084717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,4,1,power_law_1.2,0.1468160033226013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,4,1,power_law_1.2,0.16255999803543092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,4,1,power_law_1.2,0.17972480058670043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,4,1,power_law_1.2,0.16577279567718506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,4,1,power_law_1.2,0.16983040571212768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,4,power_law_1.2,12.212166595458985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,4,1,power_law_1.2,0.1685696005821228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,4,1,power_law_1.2,0.17531520128250122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,4,1,power_law_1.2,0.20769920349121093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,4,1,power_law_1.2,0.21802239418029784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,4,1,power_law_1.2,0.23447680473327637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,4,1,power_law_1.2,0.276307201385498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,4,1,power_law_1.2,0.04071039855480194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,4,1,power_law_1.01,0.8317824363708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,4,1,power_law_1.01,1.0010751724243163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,4,1,power_law_1.2,0.3266815900802612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,4,1,power_law_1.2,0.422054386138916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,4,1,power_law_1.2,0.050297600030899045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,4,1,power_law_1.2,0.05836799740791321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,4,1,power_law_1.2,0.512499189376831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,4,1,power_law_1.2,0.08386560082435608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,4,1,power_law_1.2,0.08306559920310974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,4,1,power_law_1.2,0.08910080194473266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,4,1,power_law_1.2,0.7022143840789795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,4,1,power_law_1.2,0.9038528442382813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,4,power_law_1.2,9.076787567138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,4,1,power_law_1.01,5.435500717163086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,4,1,power_law_1.2,0.06716160178184509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,4,1,power_law_1.2,0.10765440464019775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,4,1,power_law_1.2,0.07262719869613647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,4,1,power_law_1.2,0.1131327986717224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,4,1,power_law_1.2,0.0752128005027771
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,4,power_law_1.2,1.1274751663208007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,4,1,power_law_1.01,1.4698368072509767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,4,1,power_law_1.2,0.08561279773712158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,4,1,power_law_1.2,0.09199360013008118
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,4,1,power_law_1.01,1.1035264015197754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,4,balanced,0.15173866351445517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,4,balanced,0.14884266257286072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,4,balanced,0.2254506746927897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,4,balanced,0.22587200005849203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,4,balanced,0.2249493400255839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,4,balanced,0.22628267606099448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,4,balanced,0.24754667282104492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,4,balanced,0.2485919992129008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,4,balanced,0.317466676235199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,4,balanced,0.3171466588973999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,4,balanced,0.33218665917714435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,4,balanced,0.33686399459838867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,4,balanced,0.3407893180847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,4,balanced,0.35178132851918537
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,4,1,power_law_1.2,0.028384000062942505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,4,1,power_law_1.2,0.035897600650787356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,4,1,power_law_1.2,0.044352000951766966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,4,1,power_law_1.2,0.054790401458740236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,4,1,power_law_1.2,0.05502079725265503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,4,1,power_law_1.2,0.05868160128593445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,4,1,power_law_1.2,0.1226688027381897
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,4,1,power_law_1.2,0.06352639794349671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,4,1,power_law_1.2,0.12430720329284668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,4,1,power_law_1.2,0.06934400200843811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,4,1,power_law_1.2,0.09339519739151
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,4,1,power_law_1.2,0.06529279947280883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,4,1,power_law_1.2,0.10549759864807129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,4,1,power_law_1.2,0.06650879979133606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,4,1,power_law_1.2,0.11654399633407593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,4,1,power_law_1.2,0.07269120216369629
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,4,1,power_law_1.2,0.06980479955673217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,4,1,power_law_1.2,0.07379840016365051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,4,1,power_law_1.2,0.08252800107002259
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,4,1,power_law_1.2,0.08208640217781067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,4,1,power_law_1.2,0.08528640270233154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,4,1,power_law_1.2,0.09486079812049866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,4,1,power_law_1.2,0.11030399799346924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,4,1,power_law_1.2,1.2770879745483399
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,4,1,power_law_1.2,0.12540160417556762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,4,balanced,0.4052639802296956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,4,balanced,0.4156533479690552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,4,balanced,0.414250651995341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,4,balanced,0.4994239807128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,4,balanced,0.5283039808273315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,4,1,power_law_1.2,1.6739519119262696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,4,balanced,0.6809439659118652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,4,1,power_law_1.2,0.1497215986251831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,4,balanced,0.9358293215433756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,4,balanced,1.2548906803131104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,4,balanced,1.7256959279378254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,4,1,power_law_1.2,0.14318079948425294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,4,1,power_law_1.2,0.14469759464263915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,4,balanced,2.5402399698893228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,4,1,power_law_1.2,0.1697983980178833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,4,1,power_law_1.2,0.18468480110168456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,4,1,power_law_1.2,0.2113408088684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,4,balanced,3.3725334803263345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,4,1,power_law_1.2,0.21928958892822265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,4,1,power_law_1.2,0.2377471923828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,4,power_law_1.2,1.4729791641235352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,4,1,power_law_1.2,0.2582592010498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,4,balanced,5.40451176961263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,4,1,power_law_1.2,0.18172800540924072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,4,1,power_law_1.2,0.10241919755935669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,4,1,power_law_1.2,0.2987584114074707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,4,1,power_law_1.2,0.3961280107498169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,4,1,power_law_1.2,0.45357441902160645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,4,balanced,7.194762547810872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,4,balanced,0.12437867124875386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,4,balanced,0.12544533610343933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,4,balanced,0.16085333625475565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,4,balanced,0.15921599666277567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,4,balanced,0.16050666570663452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,4,balanced,0.16436800360679626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,4,balanced,0.21205333868662515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,4,balanced,0.20956265926361084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,4,balanced,0.20566399892171225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,4,balanced,0.2083466649055481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,4,balanced,0.19939732551574707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,4,balanced,0.20419732729593912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,4,balanced,0.21104532480239868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,4,balanced,0.22208533684412637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,4,1,power_law_1.2,0.6319231986999512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,4,1,power_law_1.01,1.9255168914794922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,4,1,power_law_1.2,0.8177791595458984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,4,balanced,0.15742400288581848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,4,balanced,0.15785599748293558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,4,1,power_law_1.2,0.25408639907836916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,4,balanced,0.1918506622314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,4,balanced,0.19087467590967813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,4,balanced,0.19397334257761636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,4,balanced,0.20349866151809692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,4,balanced,0.21000534296035767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,4,balanced,0.23798400163650513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,4,balanced,0.3298506736755371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,4,balanced,0.32125333944956463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,4,balanced,0.32740267117818195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,4,balanced,0.34084800879160565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,4,balanced,0.3447146813074748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,4,balanced,0.36854398250579834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,4,balanced,0.47654398282368976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,4,balanced,0.49405864874521893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,4,balanced,0.5080480178197225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,4,balanced,0.6651413440704346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,4,balanced,0.8959253629048666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,4,balanced,1.176741361618042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,4,balanced,1.5743892987569172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,4,balanced,2.2501866022745767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,4,balanced,3.0498132705688477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,4,1,power_law_1.2,0.10549119710922242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,4,1,power_law_1.2,0.10734080076217652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,4,1,power_law_1.2,0.11678719520568848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,4,balanced,4.434639930725098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,4,1,power_law_1.2,0.32315518856048586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,4,1,power_law_1.2,0.13436800241470337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,4,1,power_law_1.2,0.15067520141601562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,4,balanced,5.797375996907552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,4,balanced,0.2584106723467509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,4,balanced,0.27134400606155396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,4,balanced,0.27011199792226154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,4,balanced,0.33400531609853107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,4,balanced,0.3652533292770386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,4,balanced,0.4755520025889079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,4,1,power_law_1.2,0.1509951949119568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,4,balanced,0.6444799900054932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,4,balanced,0.8731520175933838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,4,balanced,1.1570613384246826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,4,balanced,8.683466593424479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,4,balanced,1.6998507181803386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,4,1,power_law_1.2,1.142527961730957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,4,balanced,2.234992027282715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,4,balanced,3.384677251180013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,4,balanced,11.473424275716146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,4,balanced,4.567434628804524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,4,power_law_1.01,0.15283199548721313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,4,power_law_1.01,0.1642624020576477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,4,power_law_1.01,0.2081216096878052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,4,power_law_1.01,0.22318720817565918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,4,power_law_1.01,0.22410879135131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,4,power_law_1.01,0.263372802734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,4,power_law_1.01,0.2793215990066528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,4,power_law_1.01,0.3111680030822754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,4,1,power_law_1.2,1.4875200271606446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,4,power_law_1.01,0.3915584087371826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,4,power_law_1.01,0.4011519908905029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,4,power_law_1.01,0.33656959533691405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,4,power_law_1.01,0.4128575801849365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,4,power_law_1.01,0.4208960056304932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,4,power_law_1.01,0.4460159778594971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,4,power_law_1.01,0.46062722206115725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,4,power_law_1.01,0.4848896026611328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,4,power_law_1.01,0.5876863956451416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,4,power_law_1.01,0.7098495960235596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,4,power_law_1.01,0.9164928436279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,4,1,power_law_1.2,0.17272959947586058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,4,power_law_1.01,1.45480318069458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,4,power_law_1.2,5.464364624023437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,4,power_law_1.01,0.11920000314712524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,4,power_law_1.01,0.14286719560623168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,4,power_law_1.01,0.15198080539703368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,4,power_law_1.01,0.15818239450454713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,4,1,power_law_1.2,0.20380799770355223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,4,power_law_1.01,0.15783679485321045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,4,power_law_1.01,0.19441920518875122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,4,power_law_1.01,0.2134079933166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,4,power_law_1.01,0.24067199230194092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,4,power_law_1.01,0.26550400257110596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,4,power_law_1.01,0.2744960069656372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,4,power_law_1.01,0.19949439764022828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,4,power_law_1.01,0.23032960891723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,4,power_law_1.01,0.2355072021484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,4,power_law_1.01,0.2581824064254761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,4,power_law_1.01,1.626950454711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,4,power_law_1.01,0.27777919769287107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,4,1,power_law_1.2,0.44218878746032714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,4,power_law_1.01,0.30611839294433596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,4,power_law_1.01,0.3622783899307251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,4,power_law_1.01,0.43415040969848634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,4,power_law_1.01,0.5570816040039063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,4,balanced,0.05545066793759664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,4,balanced,0.05580266813437144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,4,balanced,0.07720000048478444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,4,power_law_1.01,2.153446388244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,4,balanced,0.07784000039100647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,4,balanced,0.08107733229796092
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,4,power_law_1.2,2.06760311126709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,4,power_law_1.01,0.6620672225952149
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,4,balanced,0.08250666658083598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,4,balanced,0.0846560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,4,balanced,0.08675199747085571
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,4,balanced,0.12027200063069661
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,4,balanced,0.11788800358772278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,4,power_law_1.01,0.1627071976661682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,4,balanced,0.10925333698590596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,4,balanced,0.2840320070584615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,4,power_law_1.01,0.16933759450912475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,4,balanced,0.2775839964548747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,4,balanced,0.23458667596181235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,4,power_law_1.01,0.16879359483718873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,4,balanced,0.19455466667811075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,4,1,power_law_1.2,2.1448959350585937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,4,power_law_1.01,0.18556159734725952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,4,power_law_1.01,0.18796160221099853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,4,power_law_1.01,0.22888319492340087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,4,power_law_1.01,0.2562688112258911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,4,power_law_1.01,0.2847232103347778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,4,power_law_1.01,0.34577279090881347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,4,power_law_1.01,0.3403903961181641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,4,power_law_1.01,0.3723839998245239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,4,balanced,0.1281013290087382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,4,power_law_1.01,0.39445760250091555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,4,balanced,0.10875733693440755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,4,balanced,0.16189866264661154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,4,balanced,0.16695467631022134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,4,power_law_1.01,0.44707841873168946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,4,balanced,0.20977065960566202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,4,balanced,0.27086933453877765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,4,power_law_1.01,0.5548416137695312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,4,balanced,0.3778666655222575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,4,balanced,0.5004426638285319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,4,power_law_1.01,0.5463488101959229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,4,balanced,0.7323573430379232
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,4,balanced,0.9832159678141276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,4,power_law_1.01,0.9087552070617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,4,power_law_1.01,0.8285247802734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,4,balanced,1.4687573115030925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,4,balanced,1.9941493670145671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,4,1,power_law_1.2,0.2717632055282593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,4,1,power_law_1.2,2.4497343063354493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,4,1,power_law_1.2,0.3191488027572632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,4,power_law_1.01,0.8088704109191894
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,4,1,power_law_1.2,0.5708928108215332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,4,power_law_1.01,1.5432448387145996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,4,power_law_1.01,3.6472129821777344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,4,power_law_1.01,1.3009535789489746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,4,power_law_1.01,1.761907196044922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,4,power_law_1.01,4.683180618286133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,4,power_law_1.01,1.4941632270812988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,4,1,power_law_1.2,2.812326431274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,4,power_law_1.01,1.9849407196044921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,4,power_law_1.01,2.487455940246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,4,1,power_law_1.2,0.43438081741333007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,4,power_law_1.01,2.4816959381103514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,4,power_law_1.01,3.839923095703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,4,power_law_1.01,6.837401580810547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,4,power_law_1.01,3.7794879913330077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,4,power_law_1.01,5.386918258666992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,4,power_law_1.01,0.054022401571273804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,4,power_law_1.01,0.06723200082778931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,4,power_law_1.01,9.856473541259765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,4,power_law_1.01,0.06751360297203064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,4,power_law_1.01,0.07688959836959838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,4,power_law_1.01,0.08234239816665649
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,4,power_law_1.01,0.09859840273857116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,4,power_law_1.01,0.09120640158653259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,4,1,power_law_1.2,0.5509247779846191
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,4,power_law_1.01,0.09457280039787293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,4,power_law_1.01,0.10609279870986939
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,4,power_law_1.01,0.1004032015800476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,4,power_law_1.01,0.1171839952468872
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,4,power_law_1.01,0.1946303963661194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,4,power_law_1.01,0.15244799852371216
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,4,power_law_1.01,0.11963520050048829
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,4,power_law_1.01,0.15480320453643798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,4,power_law_1.01,0.16866559982299806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,4,power_law_1.01,0.1736448049545288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,4,power_law_1.01,0.21659519672393798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,4,power_law_1.01,0.25835518836975097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,4,power_law_1.01,0.3797440052032471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,4,power_law_1.01,7.1654914855957035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,4,power_law_1.01,0.4856832027435303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,4,power_law_1.01,0.7471551895141602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,4,power_law_1.01,14.236703491210937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,4,power_law_1.2,0.1699903964996338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,4,power_law_1.2,0.23176319599151612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,4,power_law_1.2,0.2015552043914795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,4,power_law_1.2,0.23219199180603028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,4,power_law_1.2,0.23150079250335692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,4,power_law_1.2,0.2657088041305542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,4,power_law_1.2,0.31483519077301025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,4,power_law_1.01,0.9388863563537597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,4,power_law_1.01,5.505132675170898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,4,power_law_1.2,0.31303040981292723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,4,power_law_1.2,0.3864959955215454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,4,power_law_1.2,0.4224832057952881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,4,power_law_1.2,0.3571264028549194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,4,1,power_law_1.2,4.100339126586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,4,power_law_1.2,0.3925312042236328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,4,power_law_1.2,0.39786880016326903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,4,1,power_law_1.2,0.7800191879272461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,4,power_law_1.2,0.45745282173156737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,4,power_law_1.2,0.4915071964263916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,4,power_law_1.2,0.5182015895843506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,4,power_law_1.2,0.5825215816497803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,4,power_law_1.01,1.47774076461792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,4,power_law_1.2,0.7408063888549805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,4,power_law_1.01,9.714220428466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,4,power_law_1.2,0.9084671974182129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,4,power_law_1.2,1.3170944213867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,4,1,power_law_1.2,1.019007968902588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,4,power_law_1.2,1.5532671928405761
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,4,power_law_1.01,1.8104000091552734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,4,1,power_law_1.2,0.829094409942627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,4,power_law_1.2,2.2319488525390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,4,1,power_law_1.2,3.2483646392822267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,4,power_law_1.2,0.12936320304870605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,4,power_law_1.2,0.1449280023574829
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,4,power_law_1.01,2.8554367065429687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,4,power_law_1.2,0.14599039554595947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,4,power_law_1.2,0.16373759508132935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,4,power_law_1.2,0.1760319948196411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,4,power_law_1.2,0.21944959163665773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,4,power_law_1.2,0.2614464044570923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,4,power_law_1.2,0.24442880153656005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,4,power_law_1.2,0.2718271970748901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,4,power_law_1.2,0.280185604095459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,4,power_law_1.2,0.20430080890655516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,4,power_law_1.2,0.2419584035873413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,4,power_law_1.2,0.24849920272827147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,4,power_law_1.01,17.537094116210938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,4,power_law_1.2,0.2743040084838867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,4,power_law_1.2,0.2877887964248657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,4,power_law_1.2,2.7394432067871093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,4,power_law_1.2,0.30954880714416505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,4,power_law_1.2,0.3508415937423706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,4,power_law_1.2,0.4494336128234863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,4,power_law_1.01,7.422637176513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,4,power_law_1.2,0.16554239988327027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,4,power_law_1.2,0.5624512195587158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,4,power_law_1.2,0.18202879428863525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,4,power_law_1.2,0.1838719964027405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,4,power_law_1.2,4.191513442993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,4,power_law_1.2,0.7358848094940186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,4,power_law_1.2,0.18615039587020873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,4,power_law_1.2,0.18827519416809083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,4,1,power_law_1.2,5.415852737426758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,4,power_law_1.2,0.9363712310791016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,4,power_law_1.2,0.2200063943862915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,4,power_law_1.2,0.2597631931304932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,4,power_law_1.2,0.2873471975326538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,4,1,power_law_1.2,1.4769087791442872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,4,power_law_1.2,5.697343826293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,4,power_law_1.2,1.4245887756347657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,4,power_law_1.01,19.41087951660156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,4,1,balanced,0.07019733389218648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,4,1,balanced,0.12809600432713827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,4,power_law_1.2,0.35368320941925047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,4,1,balanced,0.21241066853205362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,4,1,balanced,0.2139306664466858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,4,1,balanced,0.2137813369433085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,4,power_law_1.2,0.3492543935775757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,4,1,balanced,0.21473600467046103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,4,1,balanced,0.22787733872731528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,4,1,balanced,0.2282080054283142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,4,power_law_1.2,0.3818048000335693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,4,1,balanced,0.235642671585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,4,1,balanced,0.23734933137893677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,4,1,balanced,0.25142399470011395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,4,power_law_1.2,0.43218560218811036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,4,1,balanced,0.25409066677093506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,4,1,balanced,0.2547840078671773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,4,1,balanced,0.26291733980178833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,4,1,balanced,0.3312693238258362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,4,1,balanced,0.3340640068054199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,4,power_law_1.2,0.45754241943359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,4,1,balanced,0.3344693183898926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,4,1,balanced,0.42670400937398273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,4,1,balanced,0.4434239864349365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,4,1,balanced,0.6079626480738322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,4,1,balanced,0.7509546279907227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,4,1,balanced,1.0833546320597331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,4,1,balanced,1.3677760759989421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,4,1,balanced,2.042304039001465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,4,power_law_1.2,0.6028672218322754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,4,1,balanced,2.657621383666992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,4,power_law_1.2,0.5765632152557373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,4,1,balanced,4.153087933858235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,4,power_law_1.2,0.6782976150512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,4,1,balanced,5.655312220255534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,4,1,balanced,0.0682666649421056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,4,power_law_1.2,0.79584641456604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,4,power_law_1.01,3.685491180419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,4,1,balanced,0.07438399891058604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,4,1,balanced,0.09381866455078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,4,1,balanced,0.13382400075594583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,4,1,balanced,0.13427199920018515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,4,1,balanced,0.13369599978129068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,4,1,balanced,0.15806399782498678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,4,1,balanced,0.09483733773231506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,4,1,balanced,0.15934399763743082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,4,1,balanced,0.1316159963607788
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,4,power_law_1.2,0.05902720093727112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,4,1,balanced,0.16157333056131998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,4,1,balanced,0.130922665198644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,4,1,balanced,0.13157332936922708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,4,1,balanced,0.2676159938176473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,4,power_law_1.2,1.1990336418151855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,4,power_law_1.2,0.0683135986328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,4,1,balanced,0.2707573374112447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,4,1,balanced,0.2509866754213969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,4,power_law_1.2,0.0679423987865448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,4,1,balanced,0.24022932847340903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,4,1,balanced,0.24247467517852783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,4,power_law_1.2,1.735353660583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,4,power_law_1.2,0.07762560248374939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,4,1,balanced,0.2526560028394063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,4,1,power_law_1.2,1.1111295700073243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,4,1,balanced,0.3827626705169678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,4,1,balanced,0.3911466598510742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,4,1,balanced,0.3897706667582194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,4,1,balanced,0.5406026840209961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,4,1,balanced,0.6703680356343588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,4,1,balanced,0.9707732995351156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,4,1,balanced,0.13498133420944214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,4,1,balanced,0.15964266657829285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,4,1,balanced,1.240069309870402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,4,1,balanced,0.1602079967657725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,4,power_law_1.2,1.4432703971862793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,4,1,balanced,0.15998933712641397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,4,1,balanced,0.16156267126401266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,4,1,balanced,0.14860799908638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,4,1,balanced,0.15133866667747498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,4,1,balanced,0.15635200341542563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,4,1,balanced,0.15917866428693137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,4,1,balanced,0.200981338818868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,4,1,balanced,0.20800000429153442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,4,1,balanced,0.20030399163564047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,4,1,balanced,0.2544959982236226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,4,1,balanced,0.2683306733767192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,4,1,balanced,0.36114664872487384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,4,1,balanced,0.4394933382670085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,4,1,balanced,0.6308480103810629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,4,1,balanced,0.7976160049438477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,4,power_law_1.2,9.978387451171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,4,1,balanced,1.1817653179168701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,4,1,balanced,1.8161919911702473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,4,1,balanced,1.5162293116251628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,4,power_law_1.2,2.1880447387695314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,4,1,balanced,2.377845287322998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,4,1,balanced,2.281493345896403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,4,1,balanced,3.5078293482462564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,4,1,balanced,3.041173299153646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,4,1,power_law_1.01,0.07438719868659974
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,4,power_law_1.2,0.0894976019859314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,4,1,power_law_1.01,0.10607999563217163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,4,power_law_1.2,0.09336959719657897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,4,1,power_law_1.01,0.159443199634552
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,4,power_law_1.2,0.08831359744071961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,4,1,power_law_1.01,0.2221247911453247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,4,power_law_1.2,0.09806079864501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,4,1,power_law_1.01,0.20695040225982667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,4,1,balanced,4.492255846659343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,4,power_law_1.2,0.09240319728851318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,4,1,power_law_1.01,0.213919997215271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,4,1,balanced,0.03404266635576884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,4,1,balanced,0.055626665552457176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,4,1,balanced,0.0780213326215744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,4,1,power_law_1.01,0.23447680473327637
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,4,power_law_1.2,0.10177279710769653
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,4,1,balanced,0.0776800016562144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,4,1,balanced,0.07878399888674419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,4,1,power_law_1.01,0.2496959924697876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,4,1,balanced,0.0813920001188914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,4,power_law_1.2,0.10395519733428955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,4,1,balanced,0.09092799822489421
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,4,1,balanced,0.0939573347568512
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,4,1,balanced,0.09410666426022847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,4,power_law_1.2,0.17775360345840455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,4,1,balanced,0.0958133339881897
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,4,1,balanced,0.09831999739011128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,4,1,balanced,0.09915199875831604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,4,1,balanced,0.10129599769910176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,4,power_law_1.2,0.1780608057975769
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,4,1,balanced,0.10504532853762309
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,4,1,balanced,0.1506613294283549
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,4,1,balanced,0.13019733627637228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,4,1,balanced,0.11898666620254517
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,4,1,balanced,0.2354080080986023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,4,1,balanced,0.14830933014551798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,4,1,balanced,0.19688000281651816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,4,1,balanced,0.24468799432118735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,4,1,balanced,0.3453173240025838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,4,1,balanced,0.44727468490600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,4,1,power_law_1.01,0.2557760000228882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,4,1,balanced,0.6494773228963217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,4,1,power_law_1.01,0.25756158828735354
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,4,1,balanced,0.8549706935882568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,4,1,power_law_1.01,0.26346240043640134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,4,1,balanced,1.2501013278961182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,4,power_law_1.2,2.8201087951660155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,4,1,balanced,6.775413513183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,4,1,balanced,1.655999978383382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,4,1,power_law_1.01,0.05378559827804565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,4,1,power_law_1.01,0.06993280053138733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,4,1,power_law_1.01,0.08561279773712158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,4,1,power_law_1.01,0.130022394657135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,4,1,power_law_1.01,0.1284351944923401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,4,1,power_law_1.01,0.13559039831161498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,4,1,power_law_1.01,0.1582335948944092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,4,1,power_law_1.01,0.26988160610198975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,4,1,power_law_1.01,0.1681663990020752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,4,1,power_law_1.01,0.27352960109710694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,4,1,power_law_1.01,0.1709887981414795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,4,1,power_law_1.01,0.288319993019104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,4,1,power_law_1.01,0.18039040565490722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,4,1,power_law_1.01,0.1467519998550415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,4,1,power_law_1.01,0.31836159229278566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,4,1,power_law_1.01,0.16071679592132568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,4,1,power_law_1.01,0.16318080425262452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,4,1,power_law_1.01,0.18127360343933105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,4,power_law_1.2,0.16005120277404786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,4,1,power_law_1.01,0.19749759435653685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,4,1,balanced,8.847279866536459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,4,1,power_law_1.01,0.2049407958984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,4,power_law_1.2,0.17564799785614013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,4,1,power_law_1.01,0.07357439994812012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,4,power_law_1.2,0.1478271961212158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,4,1,power_law_1.01,0.08837119936943054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,4,1,power_law_1.01,0.22690560817718505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,4,1,power_law_1.01,0.09812480211257935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,4,1,power_law_1.01,0.12449920177459717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,4,1,power_law_1.01,0.12946560382843017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,4,power_law_1.2,2.645222473144531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,4,1,power_law_1.01,0.33724799156188967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,4,1,power_law_1.01,0.13842560052871705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,4,1,power_law_1.01,0.16092159748077392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,4,1,power_law_1.01,0.26087040901184083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,4,1,power_law_1.01,0.16899839639663697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,4,1,power_law_1.01,0.31400959491729735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,4,1,power_law_1.01,0.22561919689178467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,4,1,power_law_1.01,0.3882751941680908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,4,1,power_law_1.01,0.23621759414672852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,4,1,power_law_1.01,0.23951361179351807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,4,1,power_law_1.01,0.42917118072509763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,4,1,power_law_1.01,0.2797055959701538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,4,1,power_law_1.01,0.282425594329834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,4,power_law_1.2,0.20067200660705567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,4,1,power_law_1.01,0.3255872011184692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,4,1,power_law_1.01,0.534003210067749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,4,1,power_law_1.01,0.36830720901489256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,4,1,power_law_1.01,0.40014081001281737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,4,power_law_1.2,4.103699111938477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,4,1,power_law_1.01,0.38765439987182615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,4,power_law_1.2,0.22664320468902588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,4,1,power_law_1.01,0.46438398361206057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,4,1,power_law_1.01,0.5945663928985596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,4,1,power_law_1.01,0.48657917976379395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,4,1,power_law_1.01,0.6778240203857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,4,power_law_1.2,0.2509567975997925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,4,1,power_law_1.01,0.7236800193786621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,4,1,power_law_1.01,1.0271615982055664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,4,1,power_law_1.01,0.815334415435791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,4,power_law_1.2,0.349785590171814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,4,1,power_law_1.01,1.3003071784973144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,4,power_law_1.2,4.506585693359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,4,1,power_law_1.01,1.1577983856201173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,4,1,power_law_1.01,1.8588607788085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,4,1,power_law_1.01,0.6703872203826904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,4,1,power_law_1.01,1.4707263946533202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,4,power_law_1.2,0.5356351852416992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,4,1,power_law_1.01,2.4121599197387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,4,power_law_1.2,3.4434112548828124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,4,power_law_1.2,0.702342414855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,4,power_law_1.2,8.571897888183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,4,1,power_law_1.01,0.8480768203735352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,4,1,power_law_1.01,3.5016830444335936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,4,1,power_law_1.01,2.1046911239624024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,4,1,power_law_1.01,4.590163040161133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,4,power_law_1.2,0.9565823554992676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,4,power_law_1.2,15.445222473144531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,4,1,power_law_1.2,1.9450368881225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,4,1,power_law_1.2,0.06728320121765137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,4,1,power_law_1.2,0.10569599866867066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,4,1,power_law_1.2,0.13646080493927001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,4,1,power_law_1.2,0.20679678916931152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,4,1,power_law_1.2,0.20976641178131103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,4,1,power_law_1.2,0.20979199409484864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,4,1,power_law_1.2,0.224019193649292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,4,1,power_law_1.2,0.2348992109298706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,4,1,power_law_1.2,0.2681152105331421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,4,1,power_law_1.2,0.2659392118453979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,4,1,power_law_1.01,1.2211008071899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,4,1,power_law_1.2,0.25404160022735595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,4,1,power_law_1.2,0.2696768045425415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,4,1,power_law_1.2,0.275929594039917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,4,1,power_law_1.2,0.2850559949874878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,4,1,power_law_1.2,0.32699520587921144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,4,1,power_law_1.2,0.3306368112564087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,4,1,power_law_1.01,6.765164947509765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,4,1,power_law_1.2,0.3960832118988037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,4,1,power_law_1.2,0.43361921310424806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,4,1,power_law_1.01,0.03650560081005096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,4,1,power_law_1.01,0.04997760057449341
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,4,1,power_law_1.01,0.05775359869003296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,4,1,power_law_1.01,0.07699840068817139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,4,1,power_law_1.01,0.0758463978767395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,4,1,power_law_1.2,0.5295360088348389
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,4,1,power_law_1.01,0.08264319896697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,4,1,power_law_1.01,0.08857600092887878
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,4,1,power_law_1.01,0.09770879745483399
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,4,1,power_law_1.01,0.09088000059127807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,4,1,power_law_1.01,0.09544960260391236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,4,1,power_law_1.01,0.10304000377655029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,4,1,power_law_1.01,2.744755172729492
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,4,1,power_law_1.01,0.09912959933280945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,4,1,power_law_1.01,0.10257920026779174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,4,1,power_law_1.01,0.11751680374145508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,4,1,power_law_1.01,0.11099519729614257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,4,1,power_law_1.01,0.12353279590606689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,4,1,power_law_1.2,0.6650432109832763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,4,1,power_law_1.01,0.1373247981071472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,4,1,power_law_1.01,0.1801151990890503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,4,power_law_1.2,1.4156224250793457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,4,1,power_law_1.01,0.17875839471817018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,4,1,power_law_1.01,1.5947775840759277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,4,1,power_law_1.2,0.8151167869567871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,4,power_law_1.2,6.118630218505859
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,4,1,power_law_1.01,0.221343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,4,1,power_law_1.01,0.2744960069656372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,4,power_law_1.2,10.033267211914062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,4,1,power_law_1.2,1.143712043762207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,4,1,power_law_1.01,0.3799936056137085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,4,1,power_law_1.2,1.4553664207458497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,4,1,power_law_1.01,8.936870574951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,4,1,power_law_1.2,0.054041600227355956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,4,1,power_law_1.2,0.06792320013046264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,4,1,power_law_1.2,0.08423039913177491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,4,1,power_law_1.2,0.1253119945526123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,4,1,power_law_1.2,0.1289023995399475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,4,1,power_law_1.2,0.13964799642562867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,4,1,power_law_1.2,0.1555392026901245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,4,1,power_law_1.2,0.1713536024093628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,4,1,power_law_1.2,0.175654399394989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,4,1,power_law_1.2,0.17984639406204223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,4,1,power_law_1.2,0.14789119958877564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,4,1,power_law_1.2,0.15674879550933837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,4,1,power_law_1.2,0.1621888041496277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,4,1,power_law_1.2,0.17339520454406737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,4,1,power_law_1.2,0.1969920039176941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,4,1,power_law_1.2,0.2071615934371948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,4,1,power_law_1.2,0.2300544023513794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,4,1,power_law_1.2,0.2573247909545898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,4,power_law_1.2,1.8956544876098633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,4,1,power_law_1.2,2.0875072479248047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,4,1,power_law_1.2,0.3049855947494507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,4,1,power_law_1.01,4.019865417480469
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,4,1,power_law_1.01,0.4743040084838867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,4,1,power_law_1.2,0.39473280906677244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,4,1,power_law_1.2,0.46959362030029295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,4,1,power_law_1.01,5.447027206420898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,4,1,power_law_1.01,0.6718912124633789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,4,1,power_law_1.2,0.09177600145339966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,4,1,power_law_1.2,0.10744960308074951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,4,1,power_law_1.2,0.11679359674453735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,4,1,power_law_1.2,0.13621759414672852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,4,1,power_law_1.2,0.13675520420074463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,4,1,power_law_1.2,0.14960000514984131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,4,1,power_law_1.2,0.16716159582138063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,4,1,power_law_1.2,0.1683776021003723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,4,1,power_law_1.2,0.2271807909011841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,4,1,power_law_1.2,0.23657600879669188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,4,1,power_law_1.2,0.24817919731140137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,4,1,power_law_1.01,2.3301183700561525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,4,1,power_law_1.2,0.2800256013870239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,4,1,power_law_1.2,0.2837951898574829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,4,1,power_law_1.2,0.3043776035308838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,4,1,power_law_1.2,0.3773567914962769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,4,1,power_law_1.2,0.3861504077911377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,4,1,power_law_1.2,0.4651968002319336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,4,1,power_law_1.2,0.6612351894378662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,4,1,power_law_1.2,0.6076863765716553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,4,1,power_law_1.2,0.7195007801055908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,4,1,power_law_1.2,1.0342528343200683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,4,1,power_law_1.2,1.2983807563781737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,4,1,power_law_1.2,2.729561614990234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,4,1,power_law_1.01,0.8762240409851074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,4,1,power_law_1.01,3.0822847366333006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,4,1,power_law_1.2,1.8540159225463868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,4,power_law_1.2,2.9372095108032226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,4,power_law_1.2,16.218284606933594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,4,1,power_law_1.2,0.03459199965000152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,4,1,power_law_1.2,0.04840959906578064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,4,1,power_law_1.2,0.05589119791984558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,4,1,power_law_1.2,0.8481599807739257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,4,1,power_law_1.2,0.07633919715881347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,4,1,power_law_1.2,0.07773439884185791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,4,1,power_law_1.2,0.08217599987983704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,4,1,power_law_1.2,0.08789119720458985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,4,1,power_law_1.2,0.09672319889068604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,4,1,power_law_1.2,0.09386240243911743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,4,1,power_law_1.2,0.09409919977188111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,4,1,power_law_1.2,2.4057279586791993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,4,1,power_law_1.2,0.10254720449447632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,4,1,power_law_1.2,0.09864320158958435
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,4,1,power_law_1.2,0.1045632004737854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,4,1,power_law_1.2,0.11978240013122558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,4,1,power_law_1.2,0.115174400806427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,4,1,power_law_1.2,0.1202623963356018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,4,power_law_1.2,7.4193473815917965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,4,1,power_law_1.2,0.13405439853668213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,4,balanced,0.07439466814200084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,4,balanced,0.0983840028444926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,4,balanced,0.16403733690579733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,4,balanced,0.2634506622950236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,4,balanced,0.4426506757736206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,4,balanced,0.852346658706665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,4,balanced,0.8687893549601237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,4,1,power_law_1.2,0.15086719989776612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,4,balanced,0.8665760358174642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,4,balanced,0.969749371210734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,4,balanced,0.971781333287557
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,4,1,power_law_1.2,0.17856639623641968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,4,balanced,0.9710506598154703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,4,balanced,0.9762346744537354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,4,balanced,0.9796480337778727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,4,balanced,0.9913333257039388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,4,balanced,1.00328532854716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,4,balanced,1.0130879878997803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,4,balanced,1.0356693267822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,4,balanced,1.205509344736735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,4,balanced,1.2390080293019612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,4,balanced,1.3103093306223552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,4,balanced,1.3826826413472493
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,4,1,power_law_1.2,0.2324415922164917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,4,balanced,1.5286720593770344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,4,balanced,1.6812693277994792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,4,balanced,2.572261333465576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,4,balanced,2.8854347864786782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,4,balanced,4.1735413869222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,4,balanced,5.543178558349609
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,4,1,power_law_1.2,0.27657599449157716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,4,1,power_law_1.01,1.2717375755310059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,4,1,power_law_1.2,1.2137920379638671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,4,balanced,0.05113600194454193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,4,balanced,0.06832533578077953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,4,balanced,0.09557333588600159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,4,balanced,0.150325338045756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,4,balanced,0.26596800486246747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,4,balanced,0.490389347076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,4,balanced,0.5478826761245728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,4,balanced,0.550437331199646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,4,1,power_law_1.2,3.976563262939453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,4,balanced,0.5361280043919882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,4,1,power_law_1.2,0.38125441074371336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,4,balanced,0.5406719843546549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,4,balanced,0.5562666654586792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,4,balanced,0.5685973167419434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,4,balanced,0.5772480169932047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,4,balanced,0.5886933406194051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,4,balanced,0.6038560072580973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,4,balanced,0.6125280062357584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,4,balanced,0.636522650718689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,4,balanced,0.7514666716257731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,4,balanced,0.7841493288675944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,4,balanced,0.8560160001118978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,4,balanced,0.9286560217539469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,4,balanced,1.073957363764445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,4,balanced,1.2224213282267253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,4,balanced,1.7820587158203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,4,balanced,2.0790026982625327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,4,1,power_law_1.2,3.488710403442383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,4,balanced,3.000218709309896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,4,balanced,3.8488747278849282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,4,balanced,0.07434133191903432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,4,balanced,0.0867199997107188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,4,balanced,0.1076693336168925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,4,balanced,0.15738133589426676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,4,balanced,0.27697600920995075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,4,balanced,0.5088693300882975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,4,balanced,0.5122400124867758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,4,balanced,0.5145920117696127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,4,balanced,0.849722703297933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,4,1,power_law_1.2,1.582361602783203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,4,balanced,0.8568373521169027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,4,balanced,0.8654346466064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,4,balanced,0.872490644454956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,4,balanced,0.8836159706115723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,4,balanced,0.900218645731608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,4,balanced,0.9199519952138265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,4,balanced,0.9340960184733073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,4,balanced,0.9641866683959961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,4,balanced,1.022879997889201
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,4,1,power_law_1.2,0.4841919898986816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,4,balanced,1.0754666328430176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,4,balanced,1.1863786379496257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,4,balanced,1.2948213418324788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,4,balanced,2.0619200070699057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,4,1,power_law_1.2,5.627423858642578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,4,power_law_1.2,3.6817535400390624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,4,power_law_1.01,0.1783552050590515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,4,power_law_1.01,0.27873280048370364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,4,power_law_1.01,0.320633602142334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,4,balanced,2.2769920031229653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,4,power_law_1.01,0.4110720157623291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,4,power_law_1.01,0.5225664138793945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,4,balanced,3.2870025634765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,4,power_law_1.01,0.49358720779418946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,4,power_law_1.01,0.7599423885345459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,4,balanced,4.283429463704427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,4,power_law_1.01,0.7886144161224365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,4,power_law_1.01,0.7893631935119629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,4,power_law_1.01,0.8821120262145996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,4,balanced,6.28333854675293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,4,power_law_1.01,0.8087039947509765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,4,balanced,0.030042665700117748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,4,power_law_1.01,0.8523455619812011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,4,balanced,0.038586666186650596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,4,1,power_law_1.2,0.6789055824279785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,4,power_law_1.01,0.9032192230224609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,4,power_law_1.01,0.8940095901489258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,4,power_law_1.01,0.9145983695983887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,4,balanced,8.266101201375326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,4,power_law_1.01,0.9440511703491211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,4,balanced,0.06009600063165029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,4,power_law_1.01,0.08770560026168824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,4,balanced,0.09102933605511983
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,4,balanced,0.1446453332901001
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,4,balanced,0.2510559956232707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,4,power_law_1.01,0.15048960447311402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,4,power_law_1.01,1.0050496101379394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,4,balanced,0.2565386692682902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,4,balanced,0.25890133778254193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,4,power_law_1.01,0.19624320268630982
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,4,balanced,0.262389341990153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,4,balanced,0.26547733942667645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,4,power_law_1.01,1.1432512283325196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,4,balanced,0.2717333237330119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,4,power_law_1.01,0.22266879081726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,4,balanced,0.2670346697171529
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,4,balanced,0.27324267228444415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,4,balanced,0.28044267495473224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,4,power_law_1.01,0.2997119903564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,4,balanced,0.28070932626724243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,4,power_law_1.01,1.226201629638672
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,4,balanced,0.2852320075035095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,4,balanced,0.29583466053009033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,4,balanced,0.33581332365671795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,4,balanced,0.3264853358268738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,4,power_law_1.01,1.355519962310791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,4,balanced,0.5626826683680216
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,4,balanced,0.3991733392079671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,4,balanced,0.4182720184326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,4,power_law_1.01,1.5144960403442382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,4,balanced,0.4676320155461629
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,4,balanced,0.6785813172658285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,4,balanced,0.7746880054473877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,4,power_law_1.01,1.845292854309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,4,power_law_1.01,0.33473920822143555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,4,balanced,1.1083892981211345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,4,power_law_1.01,0.5170048236846924
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,4,balanced,1.4496159553527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,4,power_law_1.01,2.16180477142334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,4,power_law_1.01,0.4756927967071533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,4,power_law_1.01,0.47731838226318357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,4,power_law_1.01,0.4351615905761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,4,power_law_1.01,2.895827293395996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,4,power_law_1.01,0.48688640594482424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,4,power_law_1.01,0.49996161460876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,4,power_law_1.01,0.5493247985839844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,4,power_law_1.01,3.6583423614501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,4,1,power_law_1.2,0.8850175857543945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,4,power_law_1.01,0.5352191925048828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,4,power_law_1.01,0.5519743919372558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,4,power_law_1.01,0.5697599887847901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,4,power_law_1.01,0.5977663993835449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,4,power_law_1.01,0.7246848106384277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,4,power_law_1.01,0.09910399913787842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,4,power_law_1.01,0.775110387802124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,4,power_law_1.01,0.8868160247802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,4,power_law_1.01,1.0003840446472168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,4,power_law_1.01,1.2404543876647949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,4,1,power_law_1.2,4.585075378417969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,4,power_law_1.01,1.515385627746582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,4,power_law_1.01,1.933363151550293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,4,power_law_1.01,2.494963264465332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,4,power_law_1.01,0.15431679487228395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,4,power_law_1.01,0.23860480785369872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,4,power_law_1.01,3.4365440368652345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,4,power_law_1.01,0.2564671993255615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,4,power_law_1.01,0.32249600887298585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,4,power_law_1.01,4.438643264770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,4,power_law_1.01,5.1769664764404295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,4,power_law_1.01,6.432518768310547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,4,power_law_1.2,0.1576640009880066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,4,power_law_1.2,0.2513472080230713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,4,power_law_1.01,0.35957119464874265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,4,power_law_1.01,0.40536961555480955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,4,power_law_1.01,0.43201279640197754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,4,power_law_1.2,22.14593963623047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,4,power_law_1.2,0.08910080194473266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,4,power_law_1.2,0.2825216054916382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,4,power_law_1.01,0.05616000294685364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,4,power_law_1.2,0.14435839653015137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,4,power_law_1.2,0.3595328092575073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,4,power_law_1.01,0.08366079926490784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,4,power_law_1.2,0.17365119457244874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,4,power_law_1.01,0.09871360063552856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,4,power_law_1.2,0.23319680690765382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,4,power_law_1.01,0.12405120134353638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,4,power_law_1.2,0.27320959568023684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,4,power_law_1.01,0.1578559994697571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,4,power_law_1.2,0.2886143922805786
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,4,power_law_1.01,0.17329920530319215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,4,power_law_1.2,0.4276288032531738
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,4,power_law_1.01,0.21432321071624755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,4,power_law_1.2,0.47217278480529784
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,4,power_law_1.01,0.21828479766845704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,4,power_law_1.2,0.47951998710632326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,4,power_law_1.2,0.46083841323852537
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,4,power_law_1.01,0.24293758869171142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,4,1,power_law_1.2,2.3135488510131834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,4,power_law_1.2,0.43921918869018556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,4,power_law_1.01,0.22979838848114015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,4,power_law_1.2,0.5293568134307861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,4,power_law_1.2,0.4836927890777588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,4,power_law_1.2,0.7449535846710205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,4,power_law_1.2,0.5140992164611816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,4,power_law_1.2,0.5246399879455567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,4,power_law_1.2,0.7739903926849365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,4,power_law_1.2,0.5528768062591553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,4,power_law_1.2,0.8707327842712402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,4,power_law_1.2,0.5729152202606201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,4,power_law_1.2,0.7977407932281494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,4,power_law_1.2,0.5833407878875733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,4,power_law_1.01,0.24453120231628417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,4,power_law_1.2,0.8469887733459472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,4,power_law_1.2,0.6222271919250488
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,4,power_law_1.01,0.23754239082336426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,4,power_law_1.01,0.24300799369812012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,4,power_law_1.2,0.8549440383911133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,4,power_law_1.2,0.733958387374878
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,4,power_law_1.01,0.26053121089935305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,4,power_law_1.2,0.9052800178527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,4,power_law_1.2,0.819705581665039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,4,power_law_1.01,0.25795199871063235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,4,power_law_1.01,0.26874239444732667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,4,power_law_1.2,0.9156671524047851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,4,power_law_1.2,0.9315648078918457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,4,power_law_1.01,0.27603199481964114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,4,power_law_1.2,0.9390591621398926
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,4,power_law_1.01,0.30681600570678713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,4,power_law_1.2,1.0634623527526856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,4,power_law_1.01,0.759110403060913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,4,power_law_1.01,0.3307775974273682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,4,power_law_1.2,0.9547136306762696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,4,power_law_1.01,0.7734911918640137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,4,power_law_1.2,1.3151552200317382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,4,power_law_1.01,0.43911042213439944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,4,power_law_1.2,1.0013824462890626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,4,power_law_1.01,0.4152703762054443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,4,power_law_1.2,1.5758591651916505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,4,power_law_1.01,0.5102528095245361
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,4,power_law_1.01,0.6246848106384277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,4,power_law_1.2,2.03768310546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,4,power_law_1.01,0.7849599838256835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,4,power_law_1.01,0.972537612915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,4,power_law_1.01,0.7765312194824219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,4,power_law_1.2,2.589023971557617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,4,power_law_1.2,1.197209644317627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,4,power_law_1.01,0.7672128200531005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,4,power_law_1.01,1.3700096130371093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,4,power_law_1.01,0.7897151947021485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,4,power_law_1.2,1.2662272453308105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,4,power_law_1.01,0.8154815673828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,4,power_law_1.2,3.6747840881347655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,4,power_law_1.01,2.312031936645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,4,power_law_1.2,1.4567423820495606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,4,power_law_1.01,0.8467007637023926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,4,power_law_1.01,0.8881216049194336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,4,power_law_1.2,1.6127744674682618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,4,power_law_1.01,0.9687040328979493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,4,power_law_1.2,4.707878494262696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,4,power_law_1.2,0.09941120147705078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,4,power_law_1.2,1.9287872314453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,4,power_law_1.01,1.0745280265808106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,4,power_law_1.2,0.15354880094528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,4,power_law_1.2,0.18830080032348634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,4,power_law_1.01,1.1805567741394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,4,power_law_1.2,2.3350080490112304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,4,power_law_1.01,1.560051155090332
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,4,1,power_law_1.01,1.7079679489135742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,4,power_law_1.2,2.9845375061035155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,4,1,balanced,0.05686933298905691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,4,1,balanced,0.08238933483759563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,4,power_law_1.01,1.803001594543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,4,1,balanced,0.15788267056147257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,4,1,balanced,0.2568426728248596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,4,1,balanced,0.46397864818573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,4,1,balanced,0.9004373550415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,4,power_law_1.2,0.2616447925567627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,4,1,balanced,0.9001866976420084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,4,power_law_1.01,2.29749755859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,4,power_law_1.2,0.2862207889556885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,4,1,balanced,0.9034506479899088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,4,power_law_1.2,3.612268829345703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,4,1,balanced,0.9431893030802408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,4,power_law_1.2,0.3573888063430786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,4,1,balanced,0.9450453122456869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,4,1,balanced,0.93941330909729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,4,power_law_1.2,0.46660480499267576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,4,power_law_1.01,3.059334373474121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,4,1,balanced,0.9346986611684164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,4,power_law_1.2,0.44099202156066897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,4,1,balanced,0.9404533704121908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,4,1,balanced,0.9447039763132731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,4,power_law_1.2,0.7306623935699463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,4,power_law_1.2,5.391487884521484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,4,1,balanced,0.9552693367004395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,4,power_law_1.2,0.7372288227081298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,4,1,balanced,0.9633173147837321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,4,power_law_1.01,3.9287742614746093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,4,1,balanced,0.9747253259023031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,4,power_law_1.2,0.7510272026062011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,4,power_law_1.2,0.05746560096740723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,4,1,balanced,1.1909120082855225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,4,power_law_1.2,0.0845632016658783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,4,power_law_1.2,0.7806464195251465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,4,1,balanced,1.2141653696695964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,4,power_law_1.2,0.09284480214118958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,4,power_law_1.2,0.7984320163726807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,4,1,balanced,1.26090669631958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,4,power_law_1.2,0.12673920392990112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,4,1,balanced,1.3019413153330486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,4,power_law_1.2,0.7957119941711426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,4,power_law_1.2,0.1542080044746399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,4,power_law_1.2,6.535635375976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,4,power_law_1.01,5.385593414306641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,4,1,balanced,1.4087573687235515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,4,1,balanced,0.04875733455022176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,4,power_law_1.2,0.8683072090148926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,4,1,balanced,0.0642133355140686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,4,1,balanced,0.09169600407282512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,4,1,balanced,0.15679466724395752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,4,1,balanced,1.530394713083903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,4,1,balanced,0.284773329893748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,4,power_law_1.2,0.9118207931518555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,4,1,balanced,2.3930293718973794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,4,power_law_1.2,0.9817791938781738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,4,power_law_1.2,1.1477184295654297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,4,1,balanced,2.696714719136556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,4,power_law_1.2,0.18366080522537231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,4,power_law_1.01,7.377593231201172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,4,power_law_1.2,0.2168639898300171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,4,1,balanced,0.5129653215408325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,4,power_law_1.2,0.23882880210876464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,4,1,balanced,0.5893813371658325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,4,1,balanced,0.5887360175450643
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,4,power_law_1.2,0.22913920879364014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,4,1,balanced,0.5491520166397095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,4,1,balanced,0.5546720027923584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,4,power_law_1.2,0.22961280345916749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,4,1,balanced,0.5387093226114908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,4,1,balanced,0.5416213274002075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,4,power_law_1.2,0.24883201122283935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,4,1,balanced,0.5454186598459879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,4,power_law_1.2,0.24360959529876708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,4,1,balanced,0.550981322924296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,4,1,power_law_1.2,1.2725055694580079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,4,power_law_1.2,1.2696063995361329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,4,power_law_1.2,0.24226560592651367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,4,power_law_1.2,0.26356480121612547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,4,1,balanced,3.9341653188069663
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,4,power_law_1.2,0.26496639251708987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,4,power_law_1.2,1.5894399642944337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,4,power_law_1.2,0.27005441188812257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,4,power_law_1.01,9.360435485839844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,4,power_law_1.2,0.3008064031600952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,4,power_law_1.2,1.8669376373291016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,4,power_law_1.2,0.3189055919647217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,4,1,balanced,4.5845387776692705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,4,1,balanced,0.5609066486358643
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,4,power_law_1.2,0.34508800506591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,4,1,balanced,0.566981315612793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,4,1,balanced,0.5830186605453491
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,4,power_law_1.2,0.43802881240844727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,4,power_law_1.2,2.518873596191406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,4,1,balanced,0.737722635269165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,4,power_law_1.2,0.49689598083496095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,4,1,balanced,0.7577866713205973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,4,1,balanced,0.8043200174967448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,4,power_law_1.2,0.5329472064971924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,4,1,balanced,0.8513867060343424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,4,power_law_1.2,0.6751679897308349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,4,1,balanced,0.9501173496246338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,4,power_law_1.2,3.035660743713379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,4,1,balanced,1.051749308904012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,4,power_law_1.2,0.864844799041748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,4,1,balanced,1.6105440457661946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,4,power_law_1.2,1.074944019317627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,4,1,balanced,1.8088800112406414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,4,power_law_1.2,4.360518264770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,4,power_law_1.2,1.4907456398010255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,4,1,balanced,2.578848044077555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,4,1,balanced,3.384592056274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,4,power_law_1.2,2.6034048080444334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,4,1,power_law_1.01,0.056729602813720706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,4,1,power_law_1.01,0.039955198764801025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,4,power_law_1.2,5.351692962646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,4,1,power_law_1.01,0.082150399684906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,4,1,balanced,0.031162666777769726
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,4,1,balanced,0.037632000943024956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,4,1,power_law_1.01,0.058297598361968996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,4,1,balanced,0.05856533348560333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,4,1,balanced,0.08770133058230083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,4,1,power_law_1.01,0.15038080215454103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,4,1,balanced,0.1434933344523112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,4,1,power_law_1.01,0.08453119993209839
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,4,1,balanced,0.2534240086873372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,4,1,power_law_1.2,6.759584045410156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,4,1,power_law_1.01,0.2551232099533081
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,4,1,balanced,0.2581706643104553
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,4,1,balanced,0.26265066862106323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,4,1,power_law_1.01,0.14673279523849486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,4,1,balanced,0.26468799511591595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,4,1,power_law_1.01,0.3592767953872681
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,4,1,balanced,0.2696586648623149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,4,1,power_law_1.01,0.20917119979858398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,4,1,balanced,0.2750133275985718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,4,1,balanced,0.2796799937884013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,4,1,power_law_1.01,0.48279681205749514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,4,1,power_law_1.01,0.27370879650115965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,4,1,balanced,0.2860746582349141
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,4,1,balanced,0.29441599051157635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,4,1,balanced,0.3195679982503255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,4,1,power_law_1.01,0.678329610824585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,4,1,balanced,0.33100799719492596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,4,1,balanced,0.35121599833170575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,4,1,balanced,0.35953601201375324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,4,1,power_law_1.01,0.7203072071075439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,4,power_law_1.2,7.497183990478516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,4,1,balanced,0.3858453432718913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,4,1,balanced,0.4272693395614624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,4,1,power_law_1.01,0.7264319896697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,4,1,balanced,0.46832001209259033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,4,1,balanced,0.5674399932225546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,4,1,power_law_1.01,0.7501503944396972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,4,1,balanced,0.6136906544367472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,4,1,balanced,1.0843040148417156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,4,1,power_law_1.01,0.78787841796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,4,1,power_law_1.01,0.4468095779418945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,4,1,balanced,0.924618641535441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,4,1,power_law_1.01,0.7927296161651611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,4,1,power_law_1.01,0.4592383861541748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,4,1,balanced,1.3145493666330974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,4,1,power_law_1.01,0.8043007850646973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,4,1,power_law_1.01,0.43678722381591795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,4,1,power_law_1.01,0.45830402374267576
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,4,1,balanced,1.709642728169759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,4,1,power_law_1.01,0.850489616394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,4,1,balanced,0.056559999783833824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,4,1,balanced,0.07588799794514973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,4,1,power_law_1.01,0.4596672058105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,4,1,balanced,0.09872000416119893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,4,1,power_law_1.01,0.8574975967407227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,4,1,balanced,0.15312000115712485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,4,power_law_1.2,10.030169677734374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,4,1,power_law_1.01,0.46845440864562987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,4,1,power_law_1.01,0.8921855926513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,4,1,power_law_1.01,0.4795968055725098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,4,1,power_law_1.01,0.9036671638488769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,4,1,power_law_1.01,0.49025921821594237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,4,1,power_law_1.01,0.5224192142486572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,4,1,power_law_1.01,1.1160703659057618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,4,1,power_law_1.01,0.5335487842559814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,4,1,power_law_1.01,1.1681856155395507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,4,1,power_law_1.01,0.5448063850402832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,4,1,power_law_1.01,0.6994048118591308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,4,1,power_law_1.01,1.2871935844421387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,4,1,power_law_1.01,0.747430419921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,4,1,power_law_1.01,1.415839958190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,4,1,power_law_1.01,0.8384511947631836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,4,1,power_law_1.01,0.9434880256652832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,4,1,power_law_1.01,1.6737920761108398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,4,1,power_law_1.01,1.1201215744018556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,4,1,power_law_1.01,1.9070911407470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,4,1,power_law_1.01,1.2984512329101563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,4,1,power_law_1.01,0.029139199852943422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,4,1,power_law_1.01,2.467532730102539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,4,1,power_law_1.01,1.6926912307739257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,4,1,power_law_1.01,0.03689599931240082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,4,1,power_law_1.01,0.05625600218772888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,4,1,power_law_1.01,2.0627840042114256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,4,1,power_law_1.01,0.07344639897346497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,4,1,power_law_1.01,2.9709503173828127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,4,1,power_law_1.01,0.09809280037879944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,4,1,power_law_1.01,0.15177600383758544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,4,1,power_law_1.01,2.802400016784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,4,1,power_law_1.01,0.05658239722251892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,4,1,power_law_1.01,4.159564971923828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,4,1,power_law_1.01,0.08606079816818238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,4,1,power_law_1.01,3.5619583129882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,4,1,power_law_1.01,0.1162943959236145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,4,1,power_law_1.2,0.05377280116081238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,4,1,power_law_1.01,0.15947519540786742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,4,1,power_law_1.2,0.07951359748840332
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,4,1,power_law_1.01,0.20555520057678223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,4,1,balanced,0.2804373304049174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,4,1,balanced,0.5250293413798014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,4,1,power_law_1.01,0.21791999340057372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,4,1,balanced,0.5439253250757853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,4,1,power_law_1.01,5.372256088256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,4,1,power_law_1.01,0.2213184118270874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,4,1,power_law_1.2,0.04254719913005829
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,4,1,power_law_1.01,0.22648959159851073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,4,1,power_law_1.2,0.06108800172805786
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,4,1,power_law_1.01,0.2369152069091797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,4,1,power_law_1.2,0.08551679849624634
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,4,1,power_law_1.01,0.2463167905807495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,4,1,power_law_1.2,0.13814400434494017
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,4,1,power_law_1.01,0.2545023918151855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,4,1,power_law_1.2,0.15349760055541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,4,1,power_law_1.2,0.19843200445175171
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,4,1,power_law_1.01,0.27447679042816164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,4,1,power_law_1.2,0.24477438926696776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,4,1,power_law_1.2,0.25691521167755127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,4,1,power_law_1.01,0.29877119064331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,4,1,power_law_1.2,0.3211647987365723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,4,1,power_law_1.01,0.3224832057952881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,4,1,power_law_1.2,0.4298240184783936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,4,1,power_law_1.01,0.37010560035705564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,4,1,power_law_1.2,0.45226240158081055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,4,1,power_law_1.2,0.45031042098999025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,4,1,power_law_1.01,0.36205439567565917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,4,1,power_law_1.2,0.647046422958374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,4,1,power_law_1.2,0.4302847862243652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,4,1,power_law_1.01,0.42200322151184083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,4,1,power_law_1.2,0.4425151824951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,4,1,power_law_1.2,0.7036543846130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,4,1,power_law_1.01,0.43816318511962893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,4,1,power_law_1.2,0.4556096076965332
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,4,1,power_law_1.01,0.5151936054229737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,4,1,power_law_1.2,0.7332032203674317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,4,1,power_law_1.2,0.4732351779937744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,4,1,power_law_1.01,0.5573247909545899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,4,1,power_law_1.2,0.7536896228790283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,4,1,power_law_1.2,0.48514561653137206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,4,1,power_law_1.01,0.6741631984710693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,4,1,power_law_1.2,0.5027391910552979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,4,1,power_law_1.2,0.5230656147003174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,4,1,power_law_1.01,0.8382783889770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,4,1,power_law_1.2,0.5382719993591308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,4,1,power_law_1.01,1.0484288215637207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,4,1,power_law_1.2,0.5743167877197266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,4,1,power_law_1.2,0.7275263786315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,4,1,power_law_1.01,1.437606430053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,4,1,power_law_1.2,0.7840511798858643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,4,1,power_law_1.2,0.7710591793060303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,4,1,power_law_1.01,0.25048959255218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,4,1,power_law_1.2,0.8766719818115234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,4,1,power_law_1.2,0.7837952136993408
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,4,1,power_law_1.01,1.952908706665039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,4,1,power_law_1.01,0.28833279609680174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,4,1,power_law_1.2,0.852620792388916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,4,1,power_law_1.2,0.9792832374572754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,4,1,power_law_1.01,0.40471677780151366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,4,1,power_law_1.2,0.84202880859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,4,1,power_law_1.2,1.1513919830322266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,4,1,power_law_1.2,0.8903167724609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,4,1,power_law_1.2,0.9054783821105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,4,1,balanced,0.5284586747487386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,4,1,power_law_1.2,0.9239744186401367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,4,1,balanced,0.5311199824015299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,4,1,balanced,0.5325226783752441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,4,1,balanced,0.5454879999160767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,4,1,power_law_1.2,1.1440768241882324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,4,1,balanced,0.5489813486735026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,4,1,balanced,0.5516320069630941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,4,1,power_law_1.2,1.3445887565612793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,4,1,balanced,0.5502560138702393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,4,1,power_law_1.2,1.2442943572998046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,4,1,balanced,0.5595413446426392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,4,1,balanced,0.5656426747639974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,4,1,power_law_1.2,1.7306047439575196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,4,1,power_law_1.2,1.3513471603393554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,4,1,power_law_1.2,1.471008014678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,4,1,power_law_1.2,2.1018239974975588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,4,1,power_law_1.2,1.7304191589355469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,4,1,power_law_1.2,2.872083282470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,4,1,power_law_1.2,2.0085119247436523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,4,1,power_law_1.2,0.05495679974555969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,4,1,power_law_1.2,0.07312639951705932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,4,1,power_law_1.2,0.09735040068626404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,4,1,power_law_1.2,2.5018367767333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,4,1,power_law_1.2,3.619686508178711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,4,1,power_law_1.2,0.14404480457305907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,4,1,power_law_1.2,3.108780860900879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,4,1,power_law_1.2,4.22608642578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,4,1,power_law_1.01,0.4156223773956299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,4,1,power_law_1.01,0.4266816139221191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,4,1,power_law_1.01,0.4327807903289795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,4,1,power_law_1.2,5.401030349731445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,4,balanced,0.07016533116499583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,4,balanced,0.08341866731643677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,4,1,power_law_1.01,0.44670720100402833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,4,balanced,0.14261866609255472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,4,balanced,0.22579733530680338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,4,balanced,0.3867306709289551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,4,1,power_law_1.01,0.4851327896118164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,4,balanced,0.7397920290629069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,4,1,power_law_1.01,0.4922624111175537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,4,balanced,0.7483519713083903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,4,1,power_law_1.2,0.029734399914741517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,4,balanced,0.7463839848836263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,4,1,power_law_1.01,0.5282176017761231
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,4,1,power_law_1.2,0.038764798641204835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,4,balanced,0.833733320236206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,4,1,power_law_1.2,0.05798400044441223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,4,1,power_law_1.01,0.5733119964599609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,4,balanced,0.8365920384724935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,4,balanced,0.8341013590494791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,4,1,power_law_1.2,0.08293759822845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,4,1,power_law_1.01,0.5958911895751953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,4,balanced,0.8487839698791504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,4,1,power_law_1.2,0.11129599809646606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,4,balanced,0.854144016901652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,4,1,power_law_1.2,0.1503551959991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,4,1,power_law_1.2,0.2021631956100464
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,4,1,power_law_1.2,0.21482880115509034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,4,1,balanced,0.6795946756998698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,4,1,balanced,0.7116479873657227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,4,1,power_law_1.2,0.21823999881744385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,4,1,power_law_1.2,0.22353920936584473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,4,1,power_law_1.2,0.23588480949401855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,4,balanced,0.8583253224690756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,4,1,power_law_1.2,0.25014400482177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,4,balanced,0.862165371576945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,4,1,power_law_1.2,0.2604671955108643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,4,1,power_law_1.2,0.22943360805511476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,4,balanced,0.874181350072225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,4,1,power_law_1.2,0.281004810333252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,4,balanced,0.8890506426493326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,4,1,power_law_1.2,0.26484479904174807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,4,1,power_law_1.2,0.30845439434051514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,4,balanced,1.0395253499348958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,4,1,power_law_1.2,0.3980992078781128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,4,1,power_law_1.2,0.33553280830383303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,4,balanced,1.0712107022603352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,4,1,power_law_1.2,0.38407680988311765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,4,balanced,1.133951981862386
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,4,1,power_law_1.2,0.3721920013427734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,4,balanced,1.1972053050994873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,4,1,power_law_1.2,0.4336832046508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,4,balanced,1.3278453350067139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,4,1,power_law_1.2,0.45326719284057615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,4,balanced,1.4627572695414226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,4,1,power_law_1.2,0.5347583770751954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,4,balanced,2.181349277496338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,4,1,power_law_1.2,0.5749567985534668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,4,1,power_law_1.2,0.6870912075042724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,4,balanced,2.529973347981771
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,4,1,power_law_1.2,0.85665283203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,4,balanced,3.6828266779581704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,4,1,power_law_1.2,1.0576640129089356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,4,1,power_law_1.2,3.066022491455078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,4,balanced,0.04796266555786133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,4,balanced,0.06878933310508728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,4,1,power_law_1.2,1.4436800003051757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,4,balanced,0.08808533350626628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,4,balanced,0.14710399508476257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,4,balanced,0.23865600426991782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,4,balanced,0.43219200770060223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,4,1,power_law_1.2,1.709164810180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,4,balanced,0.5085279941558838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,4,balanced,0.4809066851933797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,4,balanced,0.46936531861623126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,4,1,power_law_1.2,2.125702476501465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,4,balanced,0.4713653326034546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,4,balanced,0.49081599712371826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,4,balanced,0.502618670463562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,4,balanced,0.5055093367894491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,4,balanced,0.06271466612815857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,4,balanced,0.5180533329645792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,4,balanced,0.5293546517690023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,4,1,power_law_1.2,8.964224243164063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,4,balanced,0.5389493306477865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,4,balanced,4.7349599202473955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,4,balanced,0.5618186791737875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,4,power_law_1.01,0.13147519826889037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,4,balanced,0.6634666522343954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,4,power_law_1.01,0.07996159791946411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,4,power_law_1.01,0.21699841022491456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,4,balanced,0.6949919859568278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,4,power_law_1.01,0.13772159814834595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,4,balanced,0.7599573135375977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,4,power_law_1.01,0.25160319805145265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,4,power_law_1.01,0.16961920261383057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,4,balanced,0.8251413504282633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,4,1,power_law_1.01,0.739628791809082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,4,power_law_1.01,0.34081919193267823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,4,balanced,0.9574453035990397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,4,power_law_1.01,0.2171839952468872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,4,1,power_law_1.01,0.8905023574829102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,4,power_law_1.01,0.3915456056594849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,4,balanced,1.0929653644561768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,4,power_law_1.01,0.27615358829498293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,4,power_law_1.01,0.5101696014404297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,4,power_law_1.01,0.29405438899993896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,4,balanced,1.5983039538065593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,4,power_law_1.01,0.4027071952819824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,4,power_law_1.01,0.6880000114440918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,4,balanced,1.874224026997884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,4,power_law_1.01,0.37747840881347655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,4,power_law_1.01,0.6449471950531006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,4,power_law_1.01,0.39279999732971194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,4,power_law_1.01,0.730291223526001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,4,power_law_1.01,0.3671168088912964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,4,power_law_1.01,0.720902395248413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,4,power_law_1.01,0.45186557769775393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,4,power_law_1.01,0.46036481857299805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,4,power_law_1.01,0.7125951766967773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,4,balanced,0.03158933420976003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,4,balanced,0.035546667873859406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,4,balanced,0.05801600217819214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,4,balanced,0.08362666765848796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,4,power_law_1.01,0.46030077934265134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,4,balanced,0.08040533463160197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,4,power_law_1.01,0.735200023651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,4,balanced,0.09708799918492635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,4,balanced,0.14520532886187235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,4,balanced,0.24047466119130453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,4,power_law_1.01,0.7731391906738281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,4,power_law_1.01,0.7768447875976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,4,power_law_1.01,0.798144006729126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,4,power_law_1.01,0.8083200454711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,4,balanced,0.12981866796811423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,4,balanced,0.22236265738805136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,4,balanced,0.22408000628153482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,4,power_law_1.01,0.8243647575378418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,4,power_law_1.01,0.489024019241333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,4,balanced,0.2280799945195516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,4,balanced,0.23217066129048666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,4,balanced,0.23491734266281128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,4,power_law_1.01,0.9955072402954102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,4,power_law_1.01,1.0543871879577638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,4,1,balanced,0.981818675994873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,4,power_law_1.01,1.1620991706848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,4,1,balanced,1.0470986366271973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,4,balanced,2.681152025858561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,4,power_law_1.01,1.3084927558898927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,4,power_law_1.01,0.4892864227294922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,4,balanced,0.23995733261108398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,4,balanced,0.2359093427658081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,4,power_law_1.01,0.49754881858825684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,4,balanced,0.23828266064325967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,4,1,power_law_1.2,0.4078335762023926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,4,power_law_1.01,1.5908032417297364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,4,balanced,0.24732265869776407
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,4,balanced,0.24594666560490927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,4,balanced,3.4359572728474936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,4,power_law_1.01,0.5444736003875732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,4,1,power_law_1.2,0.423199987411499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,4,balanced,0.4428853193918864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,4,power_law_1.01,0.6312064170837403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,4,balanced,0.4458826780319214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,4,1,power_law_1.2,0.43237757682800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,4,power_law_1.01,1.9280319213867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,4,balanced,0.4485013484954834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,4,power_law_1.01,0.6597951889038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,4,1,power_law_1.2,0.4521024227142334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,4,1,power_law_1.01,1.1200384140014648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,4,1,power_law_1.2,0.49249920845031736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,4,power_law_1.01,0.7863743782043457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,4,power_law_1.01,0.09048960208892823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,4,power_law_1.01,2.458803176879883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,4,1,power_law_1.2,0.5010816097259522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,4,power_law_1.01,0.14220800399780273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,4,power_law_1.01,0.8943552017211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,4,1,power_law_1.2,0.52291841506958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,4,balanced,0.2486400008201599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,4,power_law_1.01,0.1846336007118225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,4,balanced,0.2592746615409851
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,4,balanced,0.3014506697654724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,4,power_law_1.01,1.1019455909729003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,4,1,power_law_1.2,0.5772096157073975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,4,power_law_1.01,0.2273344039916992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,4,balanced,0.28682132562001544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,4,balanced,0.5029706557591757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,4,power_law_1.01,3.042950439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,4,1,power_law_1.2,0.6162559986114502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,4,power_law_1.01,0.25980799198150634
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,4,balanced,0.35102399190266925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,4,power_law_1.01,1.3300992012023927
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,4,balanced,0.3631253242492676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,4,power_law_1.01,0.32191359996795654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,4,balanced,0.4064799944559733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,4,balanced,0.5912799835205078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,4,power_law_1.01,0.3884160041809082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,4,1,power_law_1.01,1.2949888229370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,4,power_law_1.01,1.745702362060547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,4,balanced,0.6752053101857504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,4,power_law_1.01,0.3592832088470459
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,4,balanced,0.9655306339263916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,4,power_law_1.01,4.227814483642578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,4,power_law_1.01,0.6530496120452881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,4,balanced,1.256826639175415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,4,power_law_1.01,2.2265472412109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,4,power_law_1.01,0.6359360218048096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,4,power_law_1.01,0.6854656219482422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,4,power_law_1.01,0.6822720050811768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,4,power_law_1.01,3.0886783599853516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,4,power_law_1.01,0.685587215423584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,4,power_law_1.01,5.6590526580810545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,4,power_law_1.01,0.7221312046051025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,4,power_law_1.2,0.14122240543365477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,4,power_law_1.01,0.7436416149139404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,4,power_law_1.2,0.2223680019378662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,4,power_law_1.01,3.8363391876220705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,4,power_law_1.2,0.21502718925476075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,4,power_law_1.01,0.7502975940704346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,4,power_law_1.2,0.3101567983627319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,4,power_law_1.2,0.07953280210494995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,4,power_law_1.01,0.825823974609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,4,power_law_1.2,0.13646080493927001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,4,power_law_1.2,0.3916032075881958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,4,power_law_1.2,0.14654719829559326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,4,power_law_1.01,0.9560640335083008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,4,power_law_1.2,0.50698881149292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,4,power_law_1.2,0.185971200466156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,4,power_law_1.2,0.6195583820343018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,4,power_law_1.01,1.052012825012207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,4,power_law_1.2,0.2500607967376709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,4,power_law_1.2,0.5926591873168945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,4,power_law_1.01,1.2959872245788575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,4,power_law_1.2,0.6932479858398437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,4,balanced,0.7378239631652832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,4,balanced,0.7419466972351074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,4,power_law_1.2,0.7338111877441407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,4,power_law_1.01,1.5674240112304687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,4,balanced,0.7509546279907227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,4,power_law_1.01,0.05324800014495849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,4,balanced,0.7596533298492432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,4,power_law_1.2,0.7228991985321045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,4,power_law_1.01,0.07635200023651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,4,balanced,0.7682560284932455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,4,balanced,0.7848320007324219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,4,power_law_1.2,0.7563456058502197
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,4,power_law_1.01,0.09843199849128723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,4,power_law_1.2,0.25073919296264646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,4,balanced,0.7992586294809977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,4,power_law_1.01,0.11826560497283936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,4,power_law_1.2,0.7520512104034424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,4,balanced,0.8144693374633789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,4,power_law_1.2,0.4109248161315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,4,power_law_1.01,0.1433727979660034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,4,balanced,0.844650665918986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,4,power_law_1.2,0.7830976009368896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,4,power_law_1.2,0.4248191833496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,4,balanced,0.8966879844665527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,4,power_law_1.01,0.16287360191345215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,4,power_law_1.2,0.8001536369323731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,4,power_law_1.2,0.41419520378112795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,4,balanced,0.9454346497853597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,4,power_law_1.01,0.20019199848175048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,4,1,power_law_1.01,1.5276991844177246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,4,power_law_1.2,0.38173439502716067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,4,power_law_1.2,0.8030976295471192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,4,balanced,1.0480693181355794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,4,power_law_1.01,0.20594561100006104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,4,power_law_1.01,2.046444892883301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,4,power_law_1.2,0.439251184463501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,4,balanced,1.14954670270284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,4,power_law_1.2,0.8857664108276367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,4,power_law_1.2,0.4434815883636475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,4,1,power_law_1.01,1.9927167892456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,4,balanced,1.8208692868550618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,4,power_law_1.2,1.0567296028137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,4,power_law_1.2,0.4604928016662598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,4,power_law_1.01,2.5838848114013673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,4,power_law_1.2,0.47523841857910154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,4,balanced,2.0190292994181314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,4,power_law_1.2,1.1080320358276368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,4,1,power_law_1.01,2.3764928817749023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,4,power_law_1.2,0.4929344177246094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,4,power_law_1.2,0.5123456001281739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,4,power_law_1.2,1.2468671798706055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,4,balanced,2.9140427907307944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,4,1,power_law_1.2,0.7631743907928467
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,4,power_law_1.01,0.20102400779724122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,4,power_law_1.2,0.5431935787200928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,4,power_law_1.01,3.623699188232422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,4,power_law_1.01,0.21391360759735106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,4,power_law_1.2,1.403711986541748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,4,1,power_law_1.2,0.9297280311584473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,4,1,power_law_1.01,3.205363082885742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,4,power_law_1.01,0.22391679286956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,4,1,balanced,1.1129546960194905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,4,balanced,3.791722615559896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,4,power_law_1.01,0.21281280517578124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,4,power_law_1.2,1.711507225036621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,4,power_law_1.01,0.2083199977874756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,4,1,balanced,1.8695519765218098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,4,power_law_1.01,0.22899839878082276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,4,power_law_1.01,4.414931106567383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,4,power_law_1.01,0.22602880001068115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,4,power_law_1.2,2.018297576904297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,4,1,balanced,1.9643999735514324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,4,1,power_law_1.01,4.055884933471679
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,4,power_law_1.01,0.23041920661926268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,4,balanced,5.558874766031901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,4,power_law_1.2,0.6625535964965821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,4,power_law_1.01,0.2466815948486328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,4,1,balanced,2.8330666224161782
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,4,power_law_1.01,0.26460800170898435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,4,power_law_1.2,0.7172671794891358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,4,power_law_1.2,2.6728639602661133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,4,power_law_1.01,0.2925503969192505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,4,power_law_1.2,0.817574405670166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,4,power_law_1.01,0.3570175886154175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,4,1,balanced,3.7048212687174478
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,4,power_law_1.01,0.38698880672454833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,4,power_law_1.2,0.9115903854370118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,4,power_law_1.01,6.705587005615234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,4,1,power_law_1.01,5.720492935180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,4,power_law_1.2,3.1454015731811524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,4,power_law_1.01,0.4361663818359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,4,balanced,7.314512252807617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,4,power_law_1.2,1.1271936416625976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,4,power_law_1.01,0.510649585723877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,4,power_law_1.2,1.3404800415039062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,4,1,balanced,5.415589650472005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,4,power_law_1.2,0.09140480160713196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,4,power_law_1.2,4.430284881591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,4,power_law_1.2,1.8160320281982423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,4,power_law_1.2,0.14134399890899657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,4,power_law_1.2,0.17193599939346313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,4,1,power_law_1.01,7.383424377441406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,4,power_law_1.2,0.2302464008331299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,4,power_law_1.2,2.2479999542236326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,4,power_law_1.01,8.53078384399414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,4,1,balanced,7.091434478759766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,4,power_law_1.2,0.24785919189453126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,4,power_law_1.01,0.6910399913787841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,4,1,balanced,0.04964800179004669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,4,1,balanced,0.07282666862010956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,4,1,balanced,0.13446933031082153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,4,1,balanced,0.03926933308442434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,4,1,balanced,0.2243679960568746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,4,power_law_1.2,0.2594048023223877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,4,1,balanced,0.4013013442357381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,4,power_law_1.01,0.8638655662536621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,4,power_law_1.2,6.132748794555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,4,power_law_1.2,0.39779200553894045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,4,power_law_1.2,3.18668155670166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,4,power_law_1.2,0.37178239822387693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,4,power_law_1.01,1.1616640090942383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,4,power_law_1.2,0.6173952102661133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,4,power_law_1.2,0.6516543865203858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,4,1,balanced,0.051914667089780174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,4,power_law_1.01,2.1282175064086912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,4,1,balanced,0.07016533116499583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,4,1,balanced,0.09000000357627869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,4,power_law_1.2,3.9883968353271486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,4,1,balanced,0.13871467113494873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,4,power_law_1.2,0.6914688110351562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,4,1,balanced,0.7718933423360189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,4,1,power_law_1.01,0.04919039905071258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,4,1,balanced,0.7776266733805338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,4,power_law_1.2,0.669811201095581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,4,1,power_law_1.2,1.1606719970703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,4,1,power_law_1.01,0.07171199917793274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,4,1,balanced,0.7783253192901611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,4,1,power_law_1.01,0.13390079736709595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,4,power_law_1.2,0.6999551773071289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,4,1,balanced,0.8074560165405273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,4,1,power_law_1.2,1.3624256134033204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,4,power_law_1.2,0.7139200210571289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,4,power_law_1.2,0.7261631965637207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,4,1,balanced,0.055717334151268005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,4,1,balanced,0.07665599882602692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,4,1,balanced,0.13426666458447775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,4,1,balanced,0.23990933100382486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,4,power_law_1.2,0.7859968185424805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,4,1,balanced,0.433845321337382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,4,1,balanced,0.5107626517613729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,4,1,balanced,0.510261336962382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,4,power_law_1.2,0.8769984245300293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,4,1,balanced,0.4803359905878703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,4,power_law_1.2,0.0523967981338501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,4,1,balanced,0.4833279848098755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,4,1,balanced,0.46596264839172363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,4,1,balanced,0.8086133003234863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,4,1,power_law_1.01,0.22089600563049316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,4,power_law_1.2,0.9635968208312988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,4,power_law_1.2,0.07616639733314515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,4,1,balanced,0.46634666124979657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,4,1,balanced,0.8015093008677164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,4,1,balanced,0.47145601113637287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,4,power_law_1.2,0.07817599773406983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,4,1,balanced,0.4779520034790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,4,1,balanced,0.7991466522216797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,4,1,balanced,0.48412267367045086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,4,power_law_1.2,0.11026560068130493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,4,1,balanced,0.8002453645070394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,4,1,balanced,0.4910399913787842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,4,1,balanced,0.5013973315556844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,4,1,balanced,0.808730681737264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,4,power_law_1.2,0.12685439586639405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,4,1,balanced,0.6407306591669718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,4,1,balanced,0.8152533372243246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,4,1,balanced,0.030773334205150604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,4,power_law_1.2,0.1627135992050171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,4,1,balanced,0.6599306662877401
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,4,1,balanced,0.03571200122435888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,4,1,balanced,0.8216479619344076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,4,1,balanced,0.7006986935933431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,4,power_law_1.2,0.18843519687652588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,4,1,balanced,0.7416160106658936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,4,power_law_1.2,0.20383360385894775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,4,1,balanced,0.8283680280049642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,4,1,power_law_1.01,0.32027521133422854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,4,power_law_1.2,0.19407360553741454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,4,1,balanced,0.918720006942749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,4,power_law_1.2,1.119059181213379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,4,1,power_law_1.01,0.4141823768615723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,4,power_law_1.2,0.19408639669418334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,4,1,balanced,1.4072267214457195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,4,power_law_1.2,0.22933759689331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,4,1,power_law_1.01,0.577120018005371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,4,power_law_1.2,1.4016703605651855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,4,1,balanced,1.5848266283671062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,4,power_law_1.2,0.20471038818359374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,4,1,balanced,0.05409066875775655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,4,1,balanced,0.0805866668621699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,4,1,balanced,0.12800000111262003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,4,1,balanced,0.8335999647776285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,4,power_law_1.2,0.22128000259399414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,4,1,balanced,0.22177066405614218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,4,1,balanced,0.22428800662358603
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,4,power_law_1.2,0.2358783960342407
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,4,1,balanced,0.2306399941444397
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,4,1,balanced,0.2316746711730957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,4,1,balanced,0.234607994556427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,4,power_law_1.2,0.2344127893447876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,4,1,balanced,0.24009599288304648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,4,1,balanced,0.24416534105936685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,4,power_law_1.2,0.2323199987411499
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,4,1,balanced,0.24830400943756104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,4,1,balanced,0.2571786642074585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,4,power_law_1.2,0.25481600761413575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,4,1,balanced,0.282096008459727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,4,1,balanced,0.2882879972457886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,4,1,balanced,0.30878400802612305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,4,power_law_1.2,0.2868736028671265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,4,1,balanced,0.31842132409413654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,4,1,balanced,0.3380213181177775
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,4,power_law_1.2,0.3060096025466919
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,4,1,balanced,0.36985599994659424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,4,1,balanced,0.4060266812642415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,4,power_law_1.2,1.6046272277832032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,4,power_law_1.2,0.4082496166229248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,4,1,balanced,2.2653706868489585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,4,1,balanced,0.5052160024642944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,4,1,balanced,0.530784010887146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,4,power_law_1.2,0.42191362380981445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,4,1,balanced,0.24565333127975464
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,4,1,balanced,1.0020426909128826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,4,1,balanced,0.45445867379506427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,4,power_law_1.2,0.5311872005462647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,4,1,balanced,2.912245432535807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,4,power_law_1.2,2.1719488143920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,4,1,balanced,0.8037013212839762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,4,power_law_1.2,0.5659327983856202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,4,1,power_law_1.01,0.03150080144405365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,4,1,balanced,1.1367572943369548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,4,1,power_law_1.01,0.6201280117034912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,4,1,power_law_1.01,0.05363839864730835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,4,power_law_1.2,0.7696191787719726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,4,1,power_law_1.01,0.6170112133026123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,4,1,power_law_1.01,0.07538560032844543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,4,power_law_1.2,2.7055295944213866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,4,1,power_law_1.01,0.13283200263977052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,4,1,power_law_1.01,0.6398528099060059
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,4,power_law_1.2,0.8899904251098633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,4,1,power_law_1.01,0.18584959506988524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,4,1,power_law_1.01,0.6880896091461182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,4,1,power_law_1.01,0.251859188079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,4,1,balanced,0.4564319849014282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,4,power_law_1.2,1.2467583656311034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,4,power_law_1.2,3.688275146484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,4,1,power_law_1.01,0.6959167957305908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,4,1,power_law_1.01,0.6782080173492432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,4,1,balanced,1.474463939666748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,4,1,balanced,1.011082649230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,4,power_law_1.2,2.3250944137573244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,4,1,balanced,1.026527961095174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,4,1,power_law_1.01,0.7278528213500977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,4,1,balanced,1.066426674524943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,4,1,power_law_1.01,0.7508863925933837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,4,1,balanced,1.1114239692687988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,4,power_law_1.2,4.724313735961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,4,1,power_law_1.01,0.76146559715271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,4,1,balanced,1.2028213342030842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,4,1,power_law_1.01,0.3768320083618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,4,1,power_law_1.01,0.7644991874694824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,4,1,balanced,1.3010400136311848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,4,1,power_law_1.01,0.40584959983825686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,4,1,power_law_1.01,0.9598400115966796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,4,1,power_law_1.01,0.38087680339813235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,4,1,balanced,2.0006027221679688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,4,1,power_law_1.01,0.38753280639648435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,4,1,power_law_1.01,0.9980031967163085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,4,1,power_law_1.01,0.39971199035644533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,4,1,balanced,2.2511626879374185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,4,1,power_law_1.01,1.114367961883545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,4,1,power_law_1.01,0.40448641777038574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,4,power_law_1.2,6.511090850830078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,4,1,power_law_1.01,0.4131455898284912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,4,1,power_law_1.01,1.2270784378051758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,4,1,balanced,3.2422027587890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,4,1,power_law_1.01,0.4307263851165771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,4,1,power_law_1.01,0.4447296142578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,4,1,power_law_1.01,1.4460479736328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,4,1,power_law_1.01,0.4492159843444824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,4,1,power_law_1.01,0.4778111934661865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,4,1,balanced,4.255845387776692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,4,1,power_law_1.01,1.671494483947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,4,1,power_law_1.01,0.6121151924133301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,4,1,power_law_1.01,0.02839680016040802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,4,1,power_law_1.2,0.054073601961135864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,4,1,power_law_1.2,0.07872639894485474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,4,1,power_law_1.01,0.03139840066432953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,4,1,power_law_1.01,2.1328767776489257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,4,1,power_law_1.2,0.14506239891052247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,4,1,power_law_1.01,0.05330560207366943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,4,1,power_law_1.2,0.21863679885864257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,4,1,power_law_1.01,0.077811199426651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,4,power_law_1.2,9.042310333251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,4,1,power_law_1.2,0.29729280471801756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,4,1,power_law_1.01,0.10417280197143555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,4,1,power_law_1.01,2.6135168075561523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,4,1,power_law_1.2,0.031769600510597226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,4,1,power_law_1.2,0.3921855926513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,4,1,power_law_1.01,0.1414080023765564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,4,1,power_law_1.2,0.053855997323989865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,4,1,power_law_1.2,0.56910719871521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,4,1,power_law_1.2,0.07630720138549804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,4,1,power_law_1.2,1.5817407608032226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,4,1,power_law_1.01,0.6524159908294678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,4,1,power_law_1.2,0.12626559734344484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,4,1,power_law_1.2,0.5977920055389404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,4,1,power_law_1.01,0.7282623767852783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,4,1,power_law_1.2,0.17284480333328248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,4,1,power_law_1.01,3.577747344970703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,4,1,power_law_1.2,0.6212287902832031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,4,1,power_law_1.2,1.9998271942138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,4,1,power_law_1.01,0.8097344398498535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,4,1,power_law_1.2,0.22990720272064208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,4,1,power_law_1.2,0.6286399841308594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,4,1,power_law_1.2,0.38127360343933103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,4,1,power_law_1.01,0.9801983833312988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,4,1,power_law_1.2,0.6667136192321778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,4,1,power_law_1.2,0.39279999732971194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,4,1,power_law_1.01,0.17754240036010743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,4,1,balanced,0.4570399920145671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,4,1,power_law_1.2,0.6683648109436036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,4,1,power_law_1.01,1.1344639778137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,4,1,balanced,0.459882656733195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,4,1,power_law_1.2,0.37328639030456545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,4,1,power_law_1.01,4.526079940795898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,4,1,balanced,0.46116801102956134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,4,1,power_law_1.2,0.7135295867919922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,4,1,balanced,0.4660266637802124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,4,1,power_law_1.2,0.38492159843444823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,4,1,balanced,0.4692639907201131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,4,1,power_law_1.01,1.4794303894042968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,4,1,power_law_1.2,0.7402560234069824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,4,1,balanced,0.4732853174209595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,4,1,power_law_1.2,0.39971840381622314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,4,1,balanced,0.4793440103530884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,4,1,balanced,0.48554666837056476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,4,1,power_law_1.2,0.40305280685424805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,4,1,power_law_1.2,0.7652160167694092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,4,1,balanced,0.49349868297576904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,4,1,power_law_1.01,1.7970176696777345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,4,1,power_law_1.2,0.42308478355407714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,4,1,power_law_1.2,0.7747007846832276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,4,1,power_law_1.2,0.05103999972343445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,4,1,power_law_1.2,2.420582389831543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,4,1,power_law_1.2,0.4296703815460205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,4,1,power_law_1.2,0.7937088012695312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,4,1,power_law_1.01,0.187225604057312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,4,1,power_law_1.2,0.4482880115509033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,4,1,power_law_1.01,2.458579254150391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,4,1,power_law_1.01,0.04928640127182007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,4,1,power_law_1.2,0.45984640121459963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,4,1,power_law_1.2,1.013702392578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,4,1,power_law_1.01,0.06782079935073852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,4,1,power_law_1.2,0.48964481353759765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,4,1,power_law_1.2,3.2561599731445314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,4,1,power_law_1.2,1.0624704360961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,4,1,power_law_1.2,0.6304768085479736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,4,1,power_law_1.01,3.11582088470459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,4,1,power_law_1.2,1.1756608009338378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,4,1,power_law_1.2,0.6936831951141358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,4,1,power_law_1.2,0.06840320229530335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,4,1,power_law_1.2,0.7630655765533447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,4,1,power_law_1.2,0.08777599930763244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,4,1,power_law_1.2,1.2585280418395997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,4,1,power_law_1.01,0.19578880071640015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,4,1,power_law_1.2,0.12592639923095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,4,1,power_law_1.2,4.076972961425781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,4,1,power_law_1.2,0.8378687858581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,4,1,power_law_1.01,0.19827840328216553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,4,1,power_law_1.2,1.4738880157470704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,4,1,power_law_1.01,0.08750079870223999
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,4,1,power_law_1.01,0.206931209564209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,4,1,power_law_1.2,1.0029888153076172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,4,1,power_law_1.01,0.13432960510253905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,4,1,power_law_1.2,1.7172927856445312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,4,1,power_law_1.2,1.167187213897705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,4,1,power_law_1.2,1.5072575569152833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,4,1,power_law_1.2,5.748543930053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,4,1,power_law_1.2,2.1673280715942385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,4,1,power_law_1.2,1.8251327514648437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,4,1,power_law_1.01,0.21400320529937744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,4,1,power_law_1.2,2.6677696228027346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,4,1,power_law_1.01,0.22526719570159912
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,4,1,power_law_1.01,0.23970561027526854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,4,1,power_law_1.01,0.25804800987243653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,4,1,power_law_1.2,2.493177604675293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,4,1,power_law_1.2,0.02685439884662628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,4,1,power_law_1.01,0.27631359100341796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,4,1,power_law_1.2,0.030726400017738343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,4,1,power_law_1.01,0.3194943904876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,4,1,power_law_1.2,3.5604991912841797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,4,1,power_law_1.2,0.05123839974403381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,4,1,power_law_1.2,7.411885070800781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,4,1,power_law_1.01,0.3121664047241211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,4,balanced,0.02758399893840154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,4,balanced,0.02938133229811986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,4,1,power_law_1.2,0.07270399928092956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,4,balanced,0.035429333647092186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,4,balanced,0.052282666166623436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,4,balanced,0.07912000020345052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,4,1,power_law_1.2,3.138540840148926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,4,balanced,0.08057066798210144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,4,1,power_law_1.2,0.09510400295257568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,4,balanced,0.0883626639842987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,4,balanced,0.027317332724730175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,4,balanced,0.09105066458384196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,4,balanced,0.028165332973003387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,4,balanced,0.08299200236797333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,4,balanced,0.031685332457224526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,4,balanced,0.03976000100374222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,4,balanced,0.08241066833337148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,4,balanced,0.05638400216897329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,4,balanced,0.0876533289750417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,4,balanced,0.0599839985370636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,4,balanced,0.09027733405431111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,4,balanced,0.06814933319886525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,4,balanced,0.08989333113034566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,4,balanced,0.06795200208822887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,4,balanced,0.09284266829490662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,4,balanced,0.07091733316580455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,4,balanced,0.09807466467221577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,4,balanced,0.07041599849859874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,4,balanced,0.10241066416104634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,4,balanced,0.07214933137098949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,4,balanced,0.10823466380437215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,4,balanced,0.07521066566308339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,4,balanced,0.13661866386731467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,4,1,power_law_1.2,4.561471939086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,4,balanced,0.14892266194025675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,4,balanced,0.17127466201782227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,4,balanced,0.1945599913597107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,4,balanced,0.27448532978693646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,4,balanced,0.3218613266944885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,4,balanced,0.45079465707143146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,4,1,power_law_1.01,0.3623744010925293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,4,1,power_law_1.01,0.3755840063095093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,4,1,power_law_1.2,0.1295359969139099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,4,1,power_law_1.01,0.4468544006347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,4,1,power_law_1.2,0.17404799461364745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,4,balanced,0.07508266468842824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,4,1,power_law_1.01,0.48058881759643557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,4,balanced,0.08045333127180736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,4,1,power_law_1.2,0.18609919548034667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,4,balanced,0.08410666386286418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,4,balanced,0.0906773308912913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,4,1,balanced,0.5887413422266642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,4,balanced,0.09706133604049683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,4,balanced,0.11530666550000508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,4,1,power_law_1.2,0.1929792046546936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,4,1,power_law_1.01,0.5794239997863769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,4,1,balanced,0.6169013182322184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,4,1,power_law_1.2,0.19660799503326415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,4,1,balanced,0.8481600284576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,4,balanced,0.578549345334371
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,4,1,power_law_1.01,0.7157055854797363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,4,1,power_law_1.2,0.20853760242462158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,4,1,balanced,0.9081973234812418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,4,balanced,0.8305973211924235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,4,1,power_law_1.2,0.21427199840545655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,4,1,balanced,0.9662559827168783
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,4,1,power_law_1.01,0.8974080085754395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,4,1,power_law_1.2,0.22736001014709473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,4,1,balanced,1.6176692644755046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,4,1,power_law_1.2,0.19752320051193237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,4,1,power_law_1.2,0.2437376022338867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,4,1,power_law_1.2,0.26458239555358887
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,4,1,power_law_1.01,1.2316864013671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,4,1,power_law_1.2,0.23503999710083007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,4,1,balanced,1.7160479227701824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,4,1,power_law_1.2,0.2847104072570801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,4,balanced,0.12754133343696594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,4,balanced,0.15653866529464722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,4,1,power_law_1.2,0.33564159870147703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,4,balanced,0.1868799924850464
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,4,1,power_law_1.2,0.3343168020248413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,4,balanced,0.2609333395957947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,4,1,power_law_1.01,0.20838398933410646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,4,1,balanced,2.4734346071879068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,4,balanced,0.31648532549540204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,4,1,power_law_1.2,0.32245759963989257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,4,1,power_law_1.01,1.7576896667480468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,4,balanced,0.44283199310302734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,4,1,power_law_1.01,0.25333120822906496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,4,balanced,0.572927991549174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,4,balanced,1.0928320089975994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,4,balanced,0.8290080229441324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,4,1,power_law_1.01,0.34242560863494875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,4,power_law_1.01,0.031276801228523256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,4,1,balanced,3.2212851842244468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,4,power_law_1.01,0.04163840115070343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,4,balanced,1.087615966796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,4,power_law_1.01,0.04578559994697571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,4,power_law_1.01,0.029657599329948426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,4,power_law_1.01,0.052262401580810545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,4,power_law_1.01,0.03163520097732544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,4,power_law_1.01,0.05808640122413635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,4,power_law_1.01,0.03857280015945434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,4,power_law_1.01,0.07491199970245362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,4,power_law_1.01,0.04149119853973389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,4,1,balanced,4.708874702453613
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,4,balanced,0.0233599990606308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,4,power_law_1.01,0.0825984001159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,4,balanced,0.025424001117547352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,4,balanced,0.027615999182065327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,4,power_law_1.01,0.046393600106239316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,4,power_law_1.01,0.08454399704933166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,4,balanced,0.029578665892283123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,4,1,power_law_1.2,0.3760512113571167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,4,balanced,0.035786665976047516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,4,balanced,0.035616000493367515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,4,balanced,0.03708266715208689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,4,power_law_1.01,0.07822719812393189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,4,balanced,0.03566933423280716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,4,balanced,0.0379573330283165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,4,balanced,0.039664000272750854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,4,balanced,0.03998400022586187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,4,balanced,0.044495999813079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,4,power_law_1.01,0.0760640025138855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,4,balanced,0.041706666350364685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,4,1,power_law_1.2,0.3895872116088867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,4,balanced,0.040448000033696495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,4,balanced,0.04195733368396759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,4,power_law_1.01,0.08388479948043823
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,4,balanced,0.04359466830889384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,4,balanced,0.04287999868392944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,4,power_law_1.01,0.08666239976882935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,4,balanced,0.04409599800904592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,4,balanced,0.05306133131186167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,4,power_law_1.01,0.0871999979019165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,4,power_law_1.01,0.09080320000648498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,4,power_law_1.01,0.09982720017433167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,4,1,power_law_1.2,0.45989117622375486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,4,power_law_1.01,0.1090175986289978
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,4,balanced,0.051818668842315674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,4,balanced,0.051072001457214355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,4,power_law_1.01,0.11733759641647339
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,4,1,power_law_1.2,0.4903359889984131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,4,balanced,0.08003200093905131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,4,1,balanced,6.140384038289388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,4,balanced,0.06650666892528534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,4,power_law_1.01,0.13937920331954956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,4,balanced,0.07414933542410533
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,4,balanced,0.08082666496435802
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,4,1,power_law_1.2,0.5880832195281982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,4,power_law_1.01,0.15340800285339357
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,4,balanced,0.1083626647790273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,4,balanced,0.12379733721415202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,4,power_law_1.01,0.03610239923000336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,4,power_law_1.01,0.19575040340423583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,4,balanced,0.16675732533137003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,4,1,power_law_1.2,0.7385600090026856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,4,balanced,0.21041599909464517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,4,power_law_1.01,0.2310336112976074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,4,balanced,0.2935520013173421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,4,balanced,0.06193066636721293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,4,power_law_1.01,0.29487359523773193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,4,balanced,0.06497066716353099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,4,balanced,0.38339734077453613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,4,balanced,0.06462400158246358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,4,balanced,0.06574399769306183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,4,1,power_law_1.2,0.9112832069396972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,4,balanced,0.09832533200581868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,4,balanced,0.09903466701507568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,4,balanced,0.1018293301264445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,4,balanced,0.10313066840171814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,4,balanced,0.10380799571673076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,4,balanced,0.10929066936175029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,4,balanced,0.11374933520952861
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,4,1,power_law_1.2,1.2481151580810548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,4,balanced,0.12054399649302165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,4,balanced,0.12779200077056885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,4,balanced,0.14401599764823914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,4,balanced,0.16362133622169495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,4,balanced,0.23211199045181274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,4,balanced,0.27187200387318927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,4,power_law_1.01,0.053913599252700804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,4,1,power_law_1.2,1.8126399993896485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,4,power_law_1.01,0.0646336019039154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,4,power_law_1.01,0.06559360027313232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,4,power_law_1.2,0.03235200047492981
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,4,power_law_1.01,0.02476159930229187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,4,power_law_1.01,0.3726464033126831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,4,power_law_1.01,0.06427519917488098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,4,power_law_1.2,0.03883520066738129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,4,power_law_1.01,0.06643199920654297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,4,power_law_1.01,0.02656640112400055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,4,power_law_1.2,0.04451839923858643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,4,power_law_1.01,0.5055232048034668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,4,power_law_1.01,0.06915839910507202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,4,power_law_1.2,0.058195197582244874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,4,power_law_1.01,0.07223680019378662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,4,power_law_1.2,0.06349440217018128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,4,power_law_1.01,0.6456064224243164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,4,power_law_1.01,0.07370240092277527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,4,power_law_1.2,0.07541760206222534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,4,1,power_law_1.2,0.3546816110610962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,4,power_law_1.01,0.07816960215568543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,4,power_law_1.2,0.0827455997467041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,4,power_law_1.01,0.9175359725952148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,4,power_law_1.01,0.08145920038223267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,4,1,power_law_1.2,0.3624896049499512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,4,power_law_1.2,0.08568320274353028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,4,power_law_1.01,0.09052159786224365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,4,power_law_1.2,0.07742720246315002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,4,1,power_law_1.2,0.36842238903045654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,4,power_law_1.01,0.09870719909667969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,4,power_law_1.2,0.07998719811439514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,4,power_law_1.01,1.2216768264770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,4,power_law_1.01,0.1199295997619629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,4,power_law_1.2,0.08312960267066956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,4,1,power_law_1.2,0.39830400943756106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,4,power_law_1.2,0.029356798529624938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,4,power_law_1.2,0.08561919927597046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,4,1,power_law_1.2,0.41902718544006345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,4,power_law_1.2,0.08687999844551086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,4,1,power_law_1.01,0.3587007999420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,4,power_law_1.01,0.027564799785614012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,4,power_law_1.2,0.09333119988441467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,4,1,power_law_1.2,0.4309055805206299
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,4,power_law_1.01,0.029049599170684816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,4,1,power_law_1.01,0.37200639247894285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,4,power_law_1.2,0.1015552043914795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,4,1,power_law_1.2,0.4549568176269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,4,power_law_1.01,0.029926401376724244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,4,power_law_1.2,0.11175040006637574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,4,1,power_law_1.01,0.37448959350585936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,4,power_law_1.01,0.03431040048599243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,4,balanced,0.37957334518432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,4,1,power_law_1.2,0.5005504131317139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,4,power_law_1.01,0.037651199102401736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,4,1,power_law_1.01,0.38517758846282957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,4,1,power_law_1.2,0.525497579574585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,4,power_law_1.01,0.040729600191116336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,4,1,power_law_1.01,0.4175871849060059
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,4,power_law_1.01,0.039705601334571836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,4,power_law_1.01,0.13519999980926514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,4,1,power_law_1.01,0.43347840309143065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,4,power_law_1.2,0.03276160061359405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,4,power_law_1.01,0.1726207971572876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,4,power_law_1.2,0.03563520014286041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,4,1,power_law_1.01,0.4505023956298828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,4,power_law_1.01,0.20911359786987305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,4,power_law_1.2,0.04466559886932373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,4,power_law_1.01,0.27167999744415283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,4,1,power_law_1.01,0.4887551784515381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,4,power_law_1.2,0.042131200432777405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,4,power_law_1.2,0.11904640197753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,4,power_law_1.01,0.34787840843200685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,4,power_law_1.01,0.040780800580978396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,4,1,power_law_1.01,0.5139455795288086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,4,power_law_1.2,0.139302396774292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,4,balanced,0.4893813530604045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,4,power_law_1.01,0.04632959961891174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,4,power_law_1.01,0.47095680236816406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,4,power_law_1.2,0.15983359813690184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,4,balanced,0.7012906869252523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,4,power_law_1.01,0.05082880258560181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,4,power_law_1.2,0.20232961177825928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,4,power_law_1.01,0.6101247787475585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,4,power_law_1.01,0.04913919866085052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,4,power_law_1.01,0.04161919951438904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,4,power_law_1.2,0.2339008092880249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,4,power_law_1.01,0.06182399988174438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,4,power_law_1.01,0.04293760061264038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,4,power_law_1.01,0.061977601051330565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,4,power_law_1.01,0.8819135665893555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,4,power_law_1.01,0.0427264004945755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,4,power_law_1.01,0.0631168007850647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,4,power_law_1.01,0.04430719912052154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,4,power_law_1.01,0.096697598695755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,4,power_law_1.2,0.30687360763549804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,4,power_law_1.01,1.1686528205871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,4,power_law_1.01,0.04683519899845123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,4,power_law_1.2,0.054604798555374146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,4,power_law_1.2,0.37818241119384766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,4,power_law_1.01,0.04938240051269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,4,power_law_1.2,0.06307200193405152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,4,power_law_1.2,0.03498240113258362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,4,power_law_1.01,0.04906879961490631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,4,power_law_1.2,0.03966720104217529
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,4,power_law_1.01,0.0532800018787384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,4,power_law_1.2,0.045440000295639035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,4,balanced,0.9196586608886719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,4,power_law_1.01,0.06883199810981751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,4,power_law_1.2,0.04989440143108368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,4,balanced,1.3374187151590984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,4,power_law_1.01,0.07235199809074402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,4,power_law_1.2,0.05081599950790405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,4,power_law_1.01,0.08477439880371093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,4,power_law_1.2,0.06109439730644226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,4,balanced,1.7721014022827148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,4,power_law_1.2,0.06105599999427795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,4,power_law_1.01,0.09557759761810303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,4,power_law_1.2,0.062745600938797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,4,power_law_1.01,0.09660159945487976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,4,power_law_1.01,0.1186303973197937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,4,power_law_1.2,0.09711999893188476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,4,power_law_1.01,0.0986240029335022
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,4,power_law_1.01,0.14385279417037963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,4,power_law_1.2,0.09670400023460388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,4,power_law_1.01,0.10152959823608398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,4,power_law_1.2,0.0654591977596283
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,4,power_law_1.01,0.19392640590667726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,4,power_law_1.01,0.10394879579544067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,4,power_law_1.2,0.06542080044746398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,4,power_law_1.01,0.10943360328674316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,4,power_law_1.2,0.06766719818115234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,4,power_law_1.01,0.23456640243530275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,4,power_law_1.2,0.5356160163879394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,4,power_law_1.01,0.1147007942199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,4,power_law_1.01,0.33214080333709717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,4,power_law_1.01,0.12239359617233277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,4,power_law_1.01,0.13315199613571166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,4,power_law_1.2,0.023296000063419343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,4,power_law_1.01,0.4309055805206299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,4,power_law_1.01,0.16537599563598632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,4,1,balanced,0.021269333859284718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,4,1,balanced,0.02678400029738744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,4,power_law_1.2,0.02568959891796112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,4,1,balanced,0.03568000098069509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,4,1,balanced,0.05821333328882853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,4,power_law_1.01,0.19415680170059205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,4,1,balanced,0.0922986666361491
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,4,power_law_1.2,0.026950401067733765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,4,1,balanced,0.09464533130327861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,4,1,balanced,0.09985066453615825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,4,power_law_1.01,0.24670720100402832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,4,1,balanced,0.10212266445159912
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,4,power_law_1.2,0.02736639976501465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,4,1,balanced,0.08238933483759563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,4,1,balanced,0.08306133250395457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,4,power_law_1.2,0.09839360117912292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,4,1,balanced,0.0879306693871816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,4,power_law_1.2,0.029395198822021483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,4,1,balanced,0.08887466788291931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,4,1,balanced,0.08878399928410848
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,4,power_law_1.2,0.03371520042419433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,4,1,balanced,0.08845866719881694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,4,1,balanced,0.09099200367927551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,4,1,balanced,0.09467732906341553
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,4,power_law_1.2,0.037049600481987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,4,1,balanced,0.09703466296195984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,4,1,balanced,0.13182399670283
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,4,power_law_1.2,0.040249601006507874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,4,1,balanced,0.1400266687075297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,4,1,balanced,0.15562666455904642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,4,1,balanced,0.1710666616757711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,4,power_law_1.2,0.039552000164985654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,4,1,balanced,0.2400426665941874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,4,power_law_1.2,0.03973760008811951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,4,1,balanced,0.2723199923833211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,4,1,balanced,0.37588798999786377
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,4,power_law_1.2,0.0423552006483078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,4,1,balanced,0.4947093327840169
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,4,power_law_1.2,0.04307200014591217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,4,1,balanced,0.713813304901123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,4,power_law_1.2,0.045158401131629944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,4,power_law_1.01,0.3126015901565552
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,4,power_law_1.2,0.04672000110149384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,4,power_law_1.2,0.09635840058326721
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,4,power_law_1.2,0.05007359981536865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,4,power_law_1.01,0.4291967868804932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,4,power_law_1.2,0.10495359897613525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,4,power_law_1.2,0.06778879761695862
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,4,power_law_1.2,0.050432002544403075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,4,power_law_1.2,0.1076416015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,4,power_law_1.2,0.07333760261535645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,4,power_law_1.01,0.5357183933258056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,4,power_law_1.2,0.05204480290412903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,4,power_law_1.2,0.11627520322799682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,4,power_law_1.2,0.07441920042037964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,4,power_law_1.2,0.0706816017627716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,4,power_law_1.2,0.12288000583648681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,4,power_law_1.2,0.08085759878158569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,4,power_law_1.01,0.773305606842041
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,4,power_law_1.2,0.07502080202102661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,4,power_law_1.2,0.13474559783935547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,4,power_law_1.2,0.08542720079421998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,4,power_law_1.2,0.16876800060272218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,4,power_law_1.2,0.09191679954528809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,4,1,power_law_1.2,0.6663104057312011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,4,power_law_1.2,0.08650879859924317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,4,power_law_1.01,1.0044992446899415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,4,power_law_1.2,0.09959040284156799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,4,power_law_1.2,0.19540480375289918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,4,power_law_1.2,0.10380159616470337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,4,power_law_1.2,0.11890560388565063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,4,1,power_law_1.2,0.800819206237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,4,power_law_1.2,0.26270720958709715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,4,1,balanced,0.9408053557078043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,4,1,balanced,0.02120000123977661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,4,power_law_1.2,0.13793920278549193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,4,1,balanced,0.0232640008131663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,4,power_law_1.01,1.4852928161621093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,4,1,balanced,0.02834133307139079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,4,power_law_1.2,0.3284800052642822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,4,1,balanced,0.039808000127474465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,4,power_law_1.2,0.17367680072784425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,4,1,balanced,0.06218666831652323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,4,1,power_law_1.2,0.9869631767272949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,4,power_law_1.2,0.6719423770904541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,4,1,balanced,0.06427200138568878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,4,1,balanced,0.07291733225186665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,4,1,balanced,0.07379733522733052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,4,1,balanced,0.07268799841403961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,4,1,balanced,0.07147199908892314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,4,1,balanced,0.07522666454315186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,4,power_law_1.2,0.20958080291748046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,4,1,balanced,0.07787199815114339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,4,1,power_law_1.2,1.1840703964233399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,4,1,balanced,0.0766293356815974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,4,power_law_1.01,1.9266944885253907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,4,1,balanced,0.07890133559703827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,4,1,balanced,0.08275199929873149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,4,1,balanced,0.08301866551240285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,4,1,balanced,0.08880000313123067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,4,1,power_law_1.01,0.6331967830657959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,4,1,balanced,0.11660266915957133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,4,1,balanced,0.125791996717453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,4,1,power_law_1.2,1.364467239379883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,4,1,balanced,0.14290133118629456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,4,power_law_1.2,0.27487359046936033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,4,1,balanced,0.15970133741696677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,4,1,balanced,0.2304746707280477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,4,1,balanced,0.2659519910812378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,4,power_law_1.2,0.34894719123840334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,4,1,balanced,0.3689440091451009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,4,1,power_law_1.2,1.7359359741210938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,4,1,balanced,0.47969067096710205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,4,power_law_1.2,0.45457921028137205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,4,1,balanced,0.02548266698916753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,4,1,balanced,0.6919093132019043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,4,1,balanced,0.02736533433198929
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,4,1,balanced,0.02794666588306427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,4,1,balanced,0.03224000086386999
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,4,1,balanced,0.03942399968703588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,4,1,balanced,0.9149280389149984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,4,1,balanced,0.04054400076468786
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,4,1,balanced,0.04151466737190882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,4,1,power_law_1.2,2.0974143981933593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,4,1,balanced,0.04364799956480662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,4,1,power_law_1.01,0.020563200116157532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,4,1,balanced,0.043920000394185386
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,4,power_law_1.2,0.12488319873809814
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,4,1,balanced,0.04574933151404063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,4,1,balanced,0.04763199885686239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,4,1,power_law_1.01,0.026899200677871705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,4,1,balanced,0.05761066575845083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,4,power_law_1.2,0.14988160133361816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,4,1,balanced,0.058229332168896995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,4,1,power_law_1.01,0.03437440097332001
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,4,1,balanced,0.06101333101590475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,4,1,power_law_1.01,0.7783616065979004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,4,1,balanced,0.060266668597857155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,4,power_law_1.2,0.5639296054840088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,4,1,balanced,0.062037333846092224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,4,1,power_law_1.01,0.04818559885025024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,4,power_law_1.2,0.4854720115661621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,4,1,balanced,0.06572799881299336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,4,1,balanced,0.07216000060240428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,4,1,power_law_1.01,0.05729280114173889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,4,1,power_law_1.2,2.817363166809082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,4,1,balanced,0.07698133091131847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,4,1,balanced,0.0967733363310496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,4,1,power_law_1.01,0.08389760255813598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,4,1,balanced,0.10475200414657593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,4,1,balanced,0.16172800461451212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,4,1,power_law_1.01,0.08934400081634522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,4,1,balanced,0.1620319982369741
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,4,1,balanced,0.22472000122070312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,4,1,power_law_1.01,0.0964735984802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,4,1,balanced,0.2808533310890198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,4,1,power_law_1.01,0.07688959836959838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,4,1,balanced,0.38889066378275555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,4,1,power_law_1.01,0.07870720028877258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,4,1,balanced,0.5067306756973267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,4,1,power_law_1.01,0.08350719809532166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,4,1,power_law_1.2,3.5465534210205076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,4,1,power_law_1.01,0.021465599536895752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,4,1,power_law_1.01,0.08421120047569275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,4,1,power_law_1.01,0.022726400196552275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,4,1,power_law_1.01,0.0840063989162445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,4,1,power_law_1.01,0.02869119942188263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,4,1,power_law_1.01,0.08725759983062745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,4,1,power_law_1.01,0.037536001205444335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,4,1,power_law_1.01,0.08921599984169007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,4,1,power_law_1.01,0.043993601202964784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,4,1,power_law_1.01,0.09489279985427856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,4,1,power_law_1.01,0.05699840188026428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,4,1,power_law_1.01,0.11257599592208863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,4,1,power_law_1.01,0.06936320066452026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,4,1,power_law_1.01,0.13470079898834228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,4,1,power_law_1.01,0.0705024003982544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,4,1,power_law_1.01,0.1449087977409363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,4,1,power_law_1.2,5.005734252929687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,4,1,power_law_1.01,0.0692799985408783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,4,1,power_law_1.01,0.16705280542373657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,4,1,power_law_1.01,0.06959999799728393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,4,1,power_law_1.01,0.20492799282073976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,4,1,power_law_1.01,0.07215359807014465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,4,1,power_law_1.01,0.07376639842987061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,4,1,power_law_1.01,0.07339519858360291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,4,1,power_law_1.01,0.07620480060577392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,4,1,power_law_1.01,0.08027520179748535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,4,power_law_1.2,0.6067135810852051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,4,1,power_law_1.01,0.08483200073242188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,4,1,power_law_1.01,0.09786880016326904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,4,1,power_law_1.01,0.12048640251159667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,4,1,power_law_1.2,6.448831939697266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,4,1,power_law_1.01,0.13148800134658814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,4,1,power_law_1.01,0.1565440058708191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,4,1,power_law_1.01,0.25036160945892333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,4,1,power_law_1.01,0.18900480270385742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,4,1,power_law_1.01,0.30845439434051514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,4,1,power_law_1.01,0.41295361518859863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,4,1,power_law_1.01,0.5297920227050781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,4,1,power_law_1.01,0.024326400458812715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,4,1,power_law_1.01,0.7502208232879639
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,4,1,power_law_1.01,0.02475520074367523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,4,1,power_law_1.01,0.02643199861049652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,4,1,power_law_1.01,0.977836799621582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,4,1,power_law_1.01,0.029824000597000123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,4,1,power_law_1.01,0.23853440284729005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,4,1,power_law_1.2,0.020339199900627138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,4,1,power_law_1.01,0.03271040022373199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,4,1,power_law_1.2,0.02766079902648926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,4,1,power_law_1.01,0.2944960117340088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,4,1,power_law_1.01,0.03708159923553467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,4,1,power_law_1.2,0.034508800506591795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,4,1,power_law_1.01,0.03989120125770569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,4,1,power_law_1.01,0.39996159076690674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,4,1,power_law_1.2,0.043756800889968875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,4,1,power_law_1.01,0.04193919897079468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,4,1,power_law_1.2,0.05372160077095032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,4,1,power_law_1.01,0.5126463890075683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,4,1,power_law_1.01,0.04526079893112182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,4,1,power_law_1.2,0.08355200290679932
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,4,1,power_law_1.01,0.046086400747299194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,4,1,power_law_1.2,0.09266560077667237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,4,1,power_law_1.01,0.7376192092895508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,4,1,power_law_1.01,0.04962559938430786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,4,1,power_law_1.2,0.09575679898262024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,4,1,power_law_1.01,0.05103359818458557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,4,1,power_law_1.2,0.0779583990573883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,4,1,power_law_1.01,0.9638655662536622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,4,1,power_law_1.2,0.07953919768333435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,4,power_law_1.2,0.7991487979888916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,4,1,power_law_1.2,0.02152319997549057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,4,1,power_law_1.2,0.08522239923477173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,4,1,power_law_1.2,0.023155200481414794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,4,power_law_1.2,0.20807039737701416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,4,1,power_law_1.2,0.08368639945983887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,4,1,power_law_1.2,0.027923199534416198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,4,1,power_law_1.2,0.08634880185127258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,4,1,power_law_1.2,0.035692799091339114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,4,1,power_law_1.2,0.08764799833297729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,4,1,power_law_1.2,0.04012160003185272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,4,1,power_law_1.2,0.09085440039634704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,4,1,power_law_1.2,0.05818880200386047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,4,1,power_law_1.2,0.10903040170669556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,4,1,power_law_1.2,0.06961920261383056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,4,1,power_law_1.2,0.11337599754333497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,4,1,power_law_1.2,0.0711679995059967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,4,1,power_law_1.2,0.1354688048362732
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,4,1,power_law_1.01,0.05544959902763367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,4,1,power_law_1.2,0.06924160122871399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,4,1,power_law_1.2,0.1473855972290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,4,1,power_law_1.01,0.060089600086212155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,4,1,power_law_1.2,0.06993920207023621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,4,1,power_law_1.01,0.05505920052528381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,4,1,power_law_1.2,0.177183997631073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,4,1,power_law_1.2,0.07297919988632202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,4,1,power_law_1.01,0.058873599767684935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,4,1,power_law_1.2,0.07415040135383606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,4,1,power_law_1.01,0.06455039978027344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,4,1,power_law_1.2,0.07562879920005798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,4,1,power_law_1.01,0.06991999745368957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,4,1,power_law_1.2,0.07722880244255066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,4,1,power_law_1.01,0.07741439938545228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,4,1,power_law_1.2,0.08049920201301575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,4,1,power_law_1.01,0.09447680115699768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,4,1,power_law_1.2,0.08551679849624634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,4,1,power_law_1.2,0.09793279767036438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,4,1,power_law_1.01,0.11090559959411621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,4,1,power_law_1.2,0.12076159715652465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,4,1,power_law_1.01,0.14281599521636962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,4,1,power_law_1.2,0.13345919847488402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,4,1,power_law_1.01,0.1722048044204712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,4,1,power_law_1.2,0.1577407956123352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,4,1,power_law_1.01,0.23237121105194092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,4,1,power_law_1.2,0.2072767972946167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,4,1,power_law_1.01,0.2911871910095215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,4,1,power_law_1.2,0.19007999897003175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,4,1,power_law_1.2,0.23866240978240966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,4,1,power_law_1.01,0.39989120960235597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,4,1,power_law_1.2,0.2957184076309204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,4,1,power_law_1.01,0.5174335956573486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,4,power_law_1.2,0.956230354309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,4,1,power_law_1.2,0.2526655912399292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,4,1,power_law_1.2,0.30909440517425535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,4,1,power_law_1.2,0.02420479953289032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,4,1,power_law_1.01,0.9568639755249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,4,1,power_law_1.2,0.025183999538421632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,4,1,power_law_1.2,0.4023615837097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,4,1,power_law_1.2,0.0265855997800827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,4,power_law_1.2,0.8926015853881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,4,1,power_law_1.01,1.1453248023986817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,4,1,power_law_1.2,0.02892799973487854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,4,1,power_law_1.2,0.03221760094165802
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,4,1,power_law_1.2,0.036499199271202085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,4,1,power_law_1.2,0.040294399857521056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,4,1,power_law_1.2,0.04257920086383819
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,4,1,power_law_1.2,0.044889599084854126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,4,1,power_law_1.2,0.04625920057296753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,4,1,power_law_1.2,0.04935680031776428
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,4,1,power_law_1.2,0.05146239995956421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,4,1,power_law_1.2,0.417248010635376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,4,1,power_law_1.2,0.055199998617172244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,4,1,power_law_1.2,0.06007680296897888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,4,1,power_law_1.2,0.05586559772491455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,4,1,power_law_1.2,0.058841598033905027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,4,1,power_law_1.2,0.06490240097045899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,4,1,power_law_1.2,0.06986879706382751
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,4,1,power_law_1.2,0.0774399995803833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,4,1,power_law_1.2,0.09524480104446412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,4,1,power_law_1.2,0.11165440082550049
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,4,power_law_1.2,0.2540287971496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,4,1,power_law_1.2,0.14357759952545165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,4,1,power_law_1.2,0.17485439777374268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,4,1,power_law_1.2,0.5169727802276611
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,4,1,power_law_1.2,0.2339776039123535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,4,1,power_law_1.2,0.5375423908233643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,4,power_law_1.2,1.2371647834777832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,4,1,power_law_1.01,1.3227840423583985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,4,balanced,0.03787733366092046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,4,balanced,0.05574400226275126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,4,balanced,0.0791733314593633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,4,balanced,0.1288320024808248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,4,balanced,0.21418132384618124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,4,balanced,0.21724265813827515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,4,balanced,0.23079466819763184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,4,1,power_law_1.01,1.710688018798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,4,1,power_law_1.01,2.065760040283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,4,1,power_law_1.01,2.7820735931396485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,4,balanced,0.2328746716181437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,4,balanced,0.25782400369644165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,4,balanced,0.2571199933687846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,4,1,power_law_1.2,0.7397312164306641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,4,1,power_law_1.01,3.5169662475585937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,4,balanced,0.2424853245417277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,4,balanced,0.2434719999631246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,4,balanced,0.24794133504231772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,4,balanced,0.254746675491333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,4,balanced,0.2616853316624959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,4,balanced,0.2677813371022542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,4,power_law_1.2,0.3437887907028198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,4,1,power_law_1.01,4.9632831573486325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,4,balanced,0.2797813415527344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,4,balanced,0.3340533177057902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,4,balanced,0.3539786736170451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,4,balanced,0.3965919812520345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,4,power_law_1.2,1.0105600357055664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,4,balanced,0.4424479802449544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,4,balanced,0.6464426517486572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,4,balanced,0.7372213204701742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,4,balanced,1.0499253273010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,4,balanced,1.3407306671142578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,4,1,power_law_1.01,6.420269012451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,4,balanced,0.03565333286921183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,4,balanced,0.04316799839337667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,4,balanced,0.06007466713587443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,4,balanced,0.08357333143552144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,4,balanced,0.1383946637312571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,4,balanced,1.9595840771993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,4,balanced,0.1402666668097178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,4,balanced,0.15473066767056784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,4,balanced,0.15405333042144775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,4,balanced,0.15293332934379578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,4,balanced,0.15408000349998474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,4,balanced,0.15542399883270264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,4,balanced,0.1607093314329783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,4,balanced,2.556959946950277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,4,balanced,0.1665226618448893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,4,balanced,0.17270400126775107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,4,balanced,0.1789813240369161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,4,balanced,0.18961066007614136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,4,balanced,0.045647998650868736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,4,balanced,0.20485333601633707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,4,balanced,0.24476265907287598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,4,balanced,0.2696160078048706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,4,balanced,0.31523199876149494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,4,balanced,0.3632693290710449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,4,balanced,0.5216533342997233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,4,balanced,0.6161280075709025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,4,balanced,0.8745919863382975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,4,balanced,1.1247306664784749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,4,balanced,0.05239999790986379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,4,balanced,0.07034666836261749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,4,balanced,0.0885599950949351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,4,balanced,0.15037866433461508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,4,balanced,0.1556106706460317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,4,balanced,0.15618133544921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,4,balanced,0.15782399972279867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,4,balanced,0.22848532597223917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,4,balanced,0.23190933465957642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,4,balanced,0.23674132426579794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,4,balanced,0.24124266703923544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,4,balanced,0.2469759980837504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,4,balanced,0.25440533955891925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,4,balanced,0.2643466591835022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,4,balanced,0.27453867594401044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,4,balanced,0.29242666562398273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,4,balanced,0.32928532361984253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,4,balanced,0.3659733136494954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,4,balanced,1.6324106852213542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,4,balanced,0.5568853219350179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,4,balanced,0.6267306804656982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,4,balanced,2.1514612833658853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,4,balanced,0.8949759801228842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,4,balanced,1.1578453381856282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,4,balanced,1.6839040120442708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,4,1,power_law_1.2,0.29222400188446046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,4,balanced,2.2118132909139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,4,balanced,3.2420692443847656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,4,1,power_law_1.2,0.7564159870147705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,4,balanced,4.285701433817546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,4,balanced,0.029498666524887085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,4,balanced,0.02951466788848241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,4,power_law_1.01,0.0765887975692749
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,4,balanced,0.03762666632731756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,4,balanced,0.05605333546797434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,4,balanced,0.07892799874146779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,4,power_law_1.01,0.10649600028991699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,4,balanced,0.08312533299128215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,4,balanced,0.08508267005284627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,4,balanced,0.0879253347714742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,4,power_law_1.01,0.11792639493942261
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,4,balanced,0.08726400136947632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,4,balanced,0.08716799815495808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,4,power_law_1.01,0.14496639966964722
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,4,balanced,0.08876799543698628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,4,balanced,0.08961066603660583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,4,balanced,0.09166933099428813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,4,balanced,0.0926026701927185
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,4,balanced,0.11962133646011353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,4,balanced,0.1130400002002716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,4,balanced,0.10243200262387593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,4,balanced,0.18683199087778726
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,4,balanced,0.13155733545621237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,4,balanced,0.13184533516565958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,4,balanced,0.14481066664059958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,4,balanced,0.1936053236325582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,4,balanced,0.22289599974950156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,4,balanced,0.3110133409500122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,4,balanced,0.4034453233083089
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,4,balanced,0.5799146493275961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,4,power_law_1.01,0.13043839931488038
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,4,balanced,0.756341298421224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,4,power_law_1.01,0.19958399534225463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,4,power_law_1.01,0.2201535940170288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,4,power_law_1.01,0.05447040200233459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,4,power_law_1.01,0.2234368085861206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,4,power_law_1.01,0.06801279783248901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,4,power_law_1.01,0.25027201175689695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,4,power_law_1.01,0.07291520237922669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,4,power_law_1.01,0.24863998889923095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,4,power_law_1.01,0.08399999737739564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,4,power_law_1.01,0.23226239681243896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,4,power_law_1.01,0.10255359411239624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,4,1,power_law_1.2,0.9721216201782227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,4,power_law_1.01,0.2359679937362671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,4,power_law_1.01,0.1334015965461731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,4,power_law_1.01,0.24302079677581787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,4,power_law_1.01,0.1424191951751709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,4,power_law_1.01,0.25369598865509035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,4,power_law_1.01,0.1400704026222229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,4,power_law_1.01,0.255020809173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,4,power_law_1.01,0.14721920490264892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,4,power_law_1.01,0.28091518878936766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,4,power_law_1.01,0.15226880311965943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,4,power_law_1.01,0.30352001190185546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,4,power_law_1.01,0.36562559604644773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,4,power_law_1.01,0.3891072034835815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,4,power_law_1.01,0.4853055953979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,4,power_law_1.01,0.5731840133666992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,4,power_law_1.01,0.7475135803222657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,4,power_law_1.01,0.15074559450149536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,4,power_law_1.01,0.15653760433197023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,4,power_law_1.01,0.15987839698791503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,4,power_law_1.01,0.1686336040496826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,4,power_law_1.01,0.17678719758987427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,4,power_law_1.01,0.19035520553588867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,4,power_law_1.01,0.21608960628509521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,4,power_law_1.01,0.2566080093383789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,4,power_law_1.01,0.2913151979446411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,4,power_law_1.01,0.9090368270874023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,4,power_law_1.01,0.36151039600372314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,4,power_law_1.01,0.428601598739624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,4,power_law_1.01,0.06672000288963317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,4,power_law_1.01,0.07009279727935791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,4,power_law_1.01,0.5812479972839355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,4,power_law_1.01,0.08195840120315552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,4,power_law_1.01,0.7137856006622314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,4,power_law_1.01,0.10439679622650147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,4,power_law_1.01,0.10496640205383301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,4,power_law_1.01,0.9624640464782714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,4,power_law_1.01,0.14008959531784057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,4,power_law_1.01,0.14321279525756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,4,power_law_1.01,0.14782079458236694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,4,power_law_1.01,1.2662336349487304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,4,power_law_1.01,1.386342430114746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,4,power_law_1.01,0.22473599910736083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,4,power_law_1.01,0.2217855930328369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,4,power_law_1.01,0.22863359451293946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,4,power_law_1.01,1.804819107055664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,4,power_law_1.01,1.6894336700439454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,4,power_law_1.01,0.23761920928955077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,4,power_law_1.01,0.24403839111328124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,4,power_law_1.01,0.26628479957580564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,4,power_law_1.01,0.2792639970779419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,4,power_law_1.01,2.3968639373779297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,4,power_law_1.01,0.30307199954986574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,4,power_law_1.01,0.3358975887298584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,4,power_law_1.01,0.4171135902404785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,4,power_law_1.01,0.4804671764373779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,4,power_law_1.01,0.6241727828979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,4,power_law_1.01,2.435833549499512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,4,power_law_1.01,0.7953919887542724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,4,power_law_1.01,1.126150417327881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,4,power_law_1.01,3.0901695251464845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,4,power_law_1.01,1.3693504333496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,4,power_law_1.2,0.07495679855346679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,4,power_law_1.2,0.09326720237731934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,4,power_law_1.2,0.11596159934997559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,4,power_law_1.01,1.875551986694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,4,power_law_1.2,0.12949119806289672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,4,power_law_1.2,0.15488640069961548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,4,power_law_1.2,0.19454079866409302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,4,power_law_1.2,0.21682560443878174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,4,power_law_1.01,2.672275161743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,4,power_law_1.2,0.220032000541687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,4,power_law_1.2,0.25284481048583984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,4,power_law_1.2,0.25233919620513917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,4,power_law_1.2,0.2383359909057617
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,4,power_law_1.01,0.03284479975700379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,4,power_law_1.2,0.23417599201202394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,4,power_law_1.01,0.043161600828170776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,4,power_law_1.2,0.24623360633850097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,4,power_law_1.01,0.048582398891448976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,4,power_law_1.01,3.807596969604492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,4,power_law_1.2,1.5671232223510743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,4,power_law_1.2,0.2518336057662964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,4,power_law_1.01,0.059385597705841064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,4,power_law_1.2,0.2595328092575073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,4,power_law_1.01,0.060915201902389526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,4,power_law_1.2,0.2871040105819702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,4,power_law_1.01,0.07433599829673768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,4,power_law_1.2,0.3150464057922363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,4,power_law_1.01,0.0805184006690979
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,4,power_law_1.01,0.08135679960250855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,4,power_law_1.01,0.08181120157241821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,4,power_law_1.01,0.08359040021896362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,4,power_law_1.01,4.973964691162109
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,4,power_law_1.01,0.08515840172767639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,4,power_law_1.2,0.055692797899246214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,4,power_law_1.01,0.085971200466156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,4,1,power_law_1.2,0.9856767654418945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,4,power_law_1.2,0.06244480013847351
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,4,power_law_1.01,0.08712319731712341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,4,power_law_1.2,0.07486079931259156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,4,power_law_1.01,0.0929535984992981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,4,power_law_1.2,0.08151040077209473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,4,power_law_1.01,0.09868159890174866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,4,power_law_1.2,0.09953920245170593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,4,power_law_1.2,0.3656575918197632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,4,power_law_1.01,0.09953920245170593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,4,power_law_1.2,0.1331071972846985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,4,power_law_1.2,0.40956802368164064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,4,power_law_1.2,0.13848320245742798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,4,power_law_1.2,0.49996161460876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,4,power_law_1.2,0.14378880262374877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,4,power_law_1.2,1.1758975982666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,4,power_law_1.2,0.4721663951873779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,4,power_law_1.2,0.15075199604034423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,4,1,balanced,0.035216001172860466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,4,1,balanced,0.050794666012128196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,4,power_law_1.2,0.5806335926055908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,4,power_law_1.2,0.15117440223693848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,4,1,balanced,0.07074133555094402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,4,1,balanced,0.1276693344116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,4,1,balanced,0.2304533322652181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,4,power_law_1.2,0.1530943989753723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,4,1,balanced,0.2320853273073832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,4,1,balanced,0.24364266792933145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,4,power_law_1.2,0.06558079719543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,4,1,balanced,0.24496533473332724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,4,1,balanced,0.25468266010284424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,4,1,balanced,0.2537333369255066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,4,power_law_1.2,0.7831552028656006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,4,1,balanced,0.23659199476242065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,4,1,balanced,0.23589332898457846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,4,1,balanced,0.23651733001073202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,4,power_law_1.01,0.10566400289535523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,4,1,balanced,0.24089600642522177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,4,1,balanced,0.24508267641067505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,4,power_law_1.01,0.1322368025779724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,4,1,balanced,0.24909865856170654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,4,1,balanced,0.25672000646591187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,4,1,balanced,0.3202880024909973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,4,power_law_1.01,0.1382464051246643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,4,1,balanced,0.3360053300857544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,4,1,balanced,0.36375999450683594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,4,power_law_1.01,0.1506880044937134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,4,1,balanced,0.39323198795318604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,4,power_law_1.01,0.17949440479278564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,4,1,balanced,0.597274661064148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,4,power_law_1.2,0.9576767921447754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,4,power_law_1.2,0.15644160509109498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,4,1,balanced,0.6607626676559448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,4,power_law_1.01,0.21819519996643066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,4,power_law_1.2,0.06743680238723755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,4,power_law_1.2,0.16501760482788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,4,1,balanced,0.9217867056528727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,4,power_law_1.01,0.2882431983947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,4,power_law_1.2,0.08024960160255432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,4,power_law_1.2,0.17577600479125977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,4,1,balanced,1.195482651392619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,4,power_law_1.2,0.1013375997543335
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,4,power_law_1.01,0.38915839195251467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,4,power_law_1.2,0.08238080143928528
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,4,power_law_1.2,0.03276160061359405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,4,1,balanced,1.729434649149577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,4,power_law_1.2,0.14294400215148925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,4,power_law_1.2,0.03653120100498199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,4,power_law_1.2,0.14624639749526977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,4,power_law_1.2,0.04989440143108368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,4,power_law_1.2,0.14870400428771974
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,4,power_law_1.2,0.05459840297698974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,4,1,balanced,2.259119987487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,4,1,balanced,0.0273333340883255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,4,power_law_1.2,0.22576639652252198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,4,1,balanced,0.03495999922355016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,4,1,balanced,0.05366933345794678
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,4,power_law_1.2,0.059462398290634155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,4,1,power_law_1.2,0.40085759162902834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,4,power_law_1.2,0.22846078872680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,4,power_law_1.2,0.07357439994812012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,4,power_law_1.2,0.22972800731658935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,4,power_law_1.2,0.07907840013504028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,4,power_law_1.2,0.18870400190353392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,4,power_law_1.2,0.24528000354766846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,4,power_law_1.2,0.19909759759902954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,4,power_law_1.2,0.25369598865509035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,4,power_law_1.2,0.21263360977172852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,4,power_law_1.2,0.26961920261383054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,4,power_law_1.2,0.2604480028152466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,4,power_law_1.2,0.28745601177215574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,4,power_law_1.2,0.2965888023376465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,4,power_law_1.2,0.31349759101867675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,4,power_law_1.2,1.365171241760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,4,1,balanced,0.08549867073694865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,4,1,balanced,0.144186665614446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,4,power_law_1.2,0.3421119928359985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,4,1,balanced,0.14550399780273438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,4,power_law_1.2,0.3720704078674316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,4,1,balanced,0.16247999668121338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,4,1,balanced,0.16325866182645163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,4,power_law_1.2,0.4162623882293701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,4,1,balanced,0.16146666804949442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,4,1,balanced,0.16336533427238464
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,4,power_law_1.2,0.08278399705886841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,4,power_law_1.2,0.5109439849853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,4,power_law_1.2,0.08303999900817871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,4,power_law_1.2,0.45142397880554197
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,4,power_law_1.01,0.491212797164917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,4,power_law_1.2,0.08531839847564697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,4,power_law_1.2,0.6348864078521729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,4,power_law_1.2,0.5938432216644287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,4,power_law_1.2,0.08661119937896729
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,4,power_law_1.01,0.6946303844451904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,4,power_law_1.2,0.08666239976882935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,4,power_law_1.2,0.08997120261192322
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,4,power_law_1.01,1.0404159545898437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,4,power_law_1.2,0.09132800102233887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,4,power_law_1.2,0.779750394821167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,4,power_law_1.2,0.09720320105552674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,4,1,balanced,0.1570186714331309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,4,1,balanced,0.15571199854214987
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,4,power_law_1.2,0.09949439764022827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,4,power_law_1.2,1.1552895545959472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,4,power_law_1.2,0.10641280412673951
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,4,power_law_1.2,0.1627071976661682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,4,power_law_1.2,0.16210559606552125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,4,power_law_1.2,1.4435199737548827
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,4,power_law_1.2,0.1609984040260315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,4,power_law_1.2,0.18133120536804198
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,4,power_law_1.2,0.24174079895019532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,4,1,balanced,0.0378506655494372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,4,1,balanced,0.1595200002193451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,4,1,balanced,0.16499732931454977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,4,1,balanced,0.16802666584650675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,4,1,balanced,0.1722773313522339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,4,1,balanced,0.1814240018526713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,4,1,balanced,0.2303839921951294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,4,power_law_1.2,2.083302307128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,4,power_law_1.2,0.7269055843353271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,4,power_law_1.2,0.30110719203948977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,4,power_law_1.2,1.7282623291015624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,4,1,balanced,0.24787733952204385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,4,1,balanced,0.27882667382558185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,4,1,balanced,0.3096906741460164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,4,1,balanced,0.46113065878550213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,4,1,balanced,0.522277315457662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,4,1,balanced,0.7315626939137777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,4,1,balanced,0.9456693331400553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,4,1,balanced,1.3714027404785156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,4,1,balanced,1.7955093383789062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,4,power_law_1.2,0.3930624008178711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,4,1,balanced,0.02532800038655599
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,4,1,balanced,0.028736000259717304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,4,1,balanced,0.03696000079313914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,4,1,balanced,0.055306668082873024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,4,1,balanced,0.0809386670589447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,4,1,balanced,0.08309333523114522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,4,1,balanced,0.0851039985815684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,4,1,balanced,0.08663466572761536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,4,1,balanced,0.09020800391832988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,4,1,balanced,0.09215999643007915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,4,1,balanced,0.09502399961153667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,4,1,balanced,0.10388799508412679
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,4,1,balanced,0.1072746713956197
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,4,1,balanced,0.11379733681678772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,4,1,balanced,0.11265599727630615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,4,1,balanced,0.11545067032178243
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,4,1,balanced,0.12498133381207784
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,4,1,balanced,0.13270399967829385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,4,1,balanced,0.14518400033315024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,4,power_law_1.2,1.0249664306640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,4,1,balanced,0.18330132961273193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,4,1,balanced,0.18497600158055624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,4,1,balanced,0.35974399248758954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,4,1,balanced,0.2745866576830546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,4,1,balanced,0.38491201400756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,4,1,balanced,0.49272000789642334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,4,1,balanced,0.04577066500981649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,4,1,balanced,0.06234133243560791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,4,1,balanced,0.701749324798584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,4,1,balanced,0.09084799885749817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,4,1,balanced,0.916261355082194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,4,1,power_law_1.01,0.030649599432945252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,4,1,power_law_1.01,0.04715520143508911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,4,1,power_law_1.01,0.0699072003364563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,4,power_law_1.2,0.5238080024719238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,4,1,power_law_1.01,0.10964479446411132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,4,1,power_law_1.01,0.12888959646224976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,4,1,power_law_1.01,0.19782400131225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,4,1,power_law_1.01,0.22183680534362793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,4,1,power_law_1.01,0.23264000415802003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,4,1,power_law_1.01,0.24702720642089843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,4,1,power_law_1.01,0.24689919948577882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,4,1,power_law_1.01,0.22507519721984864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,4,1,power_law_1.01,0.2256511926651001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,4,1,power_law_1.01,0.22872319221496581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,4,1,power_law_1.01,0.23480958938598634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,4,1,power_law_1.01,0.24202239513397217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,4,power_law_1.2,1.3333760261535645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,4,1,power_law_1.01,0.24844799041748047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,4,1,power_law_1.01,0.28321280479431155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,4,1,power_law_1.01,0.341375994682312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,4,1,power_law_1.01,0.3780672073364258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,4,1,power_law_1.01,0.4223616123199463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,4,1,power_law_1.01,0.49441919326782224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,4,1,power_law_1.01,0.6195072174072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,4,1,power_law_1.01,0.7596159934997558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,4,1,power_law_1.01,1.0226112365722657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,4,1,power_law_1.01,1.2954815864562987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,4,1,power_law_1.01,1.835103988647461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,4,1,power_law_1.01,2.3807231903076174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,4,1,power_law_1.01,0.025824001431465148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,4,1,power_law_1.01,0.037651199102401736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,4,power_law_1.2,2.6759103775024413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,4,1,power_law_1.01,0.053446400165557864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,4,1,power_law_1.01,0.07022719979286193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,4,1,power_law_1.01,0.08588160276412964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,4,1,power_law_1.01,0.13597439527511596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,4,1,balanced,0.16036799550056458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,4,1,power_law_1.01,0.15192960500717162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,4,1,power_law_1.01,0.14791680574417115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,4,1,power_law_1.01,0.1513535976409912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,4,1,power_law_1.01,0.15282560586929322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,4,1,power_law_1.01,0.14956159591674806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,4,1,power_law_1.01,0.1504320025444031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,4,1,power_law_1.01,0.15438719987869262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,4,1,power_law_1.01,0.1597759962081909
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,4,1,power_law_1.01,0.1638592004776001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,4,1,power_law_1.01,0.1770624041557312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,4,1,power_law_1.01,0.1940608024597168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,4,1,power_law_1.01,0.24600958824157715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,4,1,power_law_1.01,0.2737663984298706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,4,1,power_law_1.01,0.3160896062850952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,4,1,power_law_1.01,0.36986238956451417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,4,1,power_law_1.01,0.47260799407958987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,4,1,power_law_1.01,0.578707218170166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,4,1,power_law_1.01,0.7903808116912842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,4,1,power_law_1.01,1.0050559997558595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,4,power_law_1.2,2.013471984863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,4,1,power_law_1.01,1.4386560440063476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,4,power_law_1.2,0.7302464008331299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,4,1,power_law_1.01,0.03583360016345978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,4,1,power_law_1.01,0.044409599900245664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,4,1,power_law_1.01,0.06104320287704468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,4,1,power_law_1.01,1.8616960525512696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,4,1,power_law_1.01,0.0782912015914917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,4,1,power_law_1.01,0.09141759872436524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,4,1,power_law_1.01,0.14426239728927612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,4,1,power_law_1.01,0.14853760004043579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,4,1,power_law_1.01,0.15077120065689087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,4,1,power_law_1.01,0.1548416018486023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,4,power_law_1.2,2.424870491027832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,4,1,power_law_1.01,0.1574336051940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,4,1,power_law_1.01,0.16468479633331298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,4,1,power_law_1.01,0.1831231951713562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,4,1,power_law_1.01,0.19160959720611573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,4,1,power_law_1.01,0.19472639560699462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,4,1,power_law_1.01,0.21264638900756835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,4,1,power_law_1.01,0.024505600333213806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,4,1,power_law_1.01,0.2377216100692749
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,4,1,power_law_1.01,0.027795198559761047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,4,1,power_law_1.01,0.31342079639434817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,4,1,power_law_1.01,0.0366784006357193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,4,1,power_law_1.01,0.3482687950134277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,4,1,power_law_1.01,0.04780159890651703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,4,1,power_law_1.01,0.05926399827003479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,4,1,power_law_1.01,0.40988798141479493
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,4,1,power_law_1.01,0.07339519858360291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,4,1,balanced,0.15873066584269205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,4,1,power_law_1.01,0.5097280025482178
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,4,1,power_law_1.01,0.07847679853439331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,4,1,balanced,0.16014933586120605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,4,1,balanced,0.15992533167203268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,4,1,balanced,0.16216533382733664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,4,1,balanced,0.16085333625475565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,4,1,power_law_1.01,0.6102335929870606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,4,1,balanced,0.1655946671962738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,4,1,balanced,0.1662613352139791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,4,1,balanced,0.167738676071167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,4,1,balanced,0.17756267388661703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,4,1,balanced,0.22848000129063925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,4,1,balanced,0.23162666956583658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,4,1,balanced,0.27246934175491333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,4,1,balanced,0.29049066702524823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,4,1,balanced,0.3110186656316121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,4,1,balanced,0.5013866821924845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,4,1,balanced,0.5384853283564249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,4,1,power_law_1.01,0.08204159736633301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,4,1,power_law_1.01,0.08621439933776856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,4,1,power_law_1.01,0.08716800212860107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,4,1,power_law_1.01,0.0934656023979187
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,4,1,power_law_1.01,0.0982591986656189
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,4,1,power_law_1.01,0.1051967978477478
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,4,1,power_law_1.01,0.11706240177154541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,4,1,power_law_1.01,0.10763520002365112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,4,1,power_law_1.01,0.11296000480651855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,4,1,power_law_1.01,0.12819199562072753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,4,1,power_law_1.01,0.1323904037475586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,4,1,power_law_1.01,0.15342719554901124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,4,1,power_law_1.01,0.1658560037612915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,4,1,power_law_1.01,0.1965824007987976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,4,1,power_law_1.01,0.2473088026046753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,4,1,power_law_1.01,0.30078079700469973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,4,1,power_law_1.01,0.4058688163757324
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,4,1,power_law_1.01,0.5200191974639893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,4,1,power_law_1.01,0.73023362159729
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,4,1,power_law_1.01,0.9861248016357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,4,1,power_law_1.2,0.030272001028060914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,4,1,power_law_1.2,0.0471807986497879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,4,1,power_law_1.2,0.06924800276756286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,4,1,power_law_1.2,0.10308480262756348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,4,1,power_law_1.2,0.12272640466690063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,4,1,power_law_1.2,0.1993280053138733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,4,1,power_law_1.2,0.22645120620727538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,4,1,power_law_1.2,0.22853119373321534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,4,1,power_law_1.2,0.2439039945602417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,4,1,power_law_1.2,0.2513344049453735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,4,1,power_law_1.2,0.2300800085067749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,4,1,power_law_1.2,0.22848000526428222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,4,1,power_law_1.2,0.2337536096572876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,4,1,power_law_1.2,0.23578240871429443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,4,1,power_law_1.2,0.2405695915222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,4,1,power_law_1.2,0.2535423994064331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,4,1,power_law_1.2,0.2823040008544922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,4,1,power_law_1.2,0.36029438972473143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,4,1,power_law_1.2,0.38216960430145264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,4,1,power_law_1.2,0.43837437629699705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,4,1,power_law_1.2,0.5017856121063232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,4,1,power_law_1.2,0.6285120010375976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,4,power_law_1.2,3.90447998046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,4,1,power_law_1.2,0.5256576061248779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,4,1,power_law_1.2,0.026476800441741943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,4,1,power_law_1.2,0.03689599931240082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,4,1,power_law_1.2,0.05025920271873474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,4,1,power_law_1.2,0.06481919884681701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,4,1,power_law_1.2,0.08239359855651855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,4,1,power_law_1.2,0.1358464002609253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,4,1,power_law_1.2,0.1496448040008545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,4,1,power_law_1.2,0.1533504009246826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,4,1,power_law_1.2,0.7768256187438964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,4,1,power_law_1.2,0.15324800014495848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,4,1,power_law_1.2,0.1560960054397583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,4,1,power_law_1.2,1.0337920188903809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,4,1,power_law_1.2,0.15075839757919313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,4,1,power_law_1.2,0.1525823950767517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,4,1,power_law_1.2,0.1581760048866272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,4,1,power_law_1.2,0.16310399770736694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,4,1,power_law_1.01,0.8301440238952636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,4,1,power_law_1.2,0.16809600591659546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,4,1,power_law_1.2,0.17823359966278077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,4,1,balanced,0.7429920037587484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,4,1,power_law_1.01,1.0417216300964356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,4,1,power_law_1.2,0.1983423948287964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,4,1,balanced,0.9742720127105713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,4,1,power_law_1.2,0.2549312114715576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,4,1,power_law_1.2,0.27658240795135497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,4,1,balanced,1.4040320714314778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,4,1,power_law_1.01,1.4556991577148437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,4,1,power_law_1.2,0.3205312013626099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,4,1,balanced,1.8441333770751953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,4,1,power_law_1.2,0.37861759662628175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,4,1,power_law_1.01,1.874131202697754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,4,1,power_law_1.2,0.47859840393066405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,4,1,balanced,2.6899360020955405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,4,1,power_law_1.01,2.7042047500610353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,4,1,balanced,3.5753227869669595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,4,1,power_law_1.2,0.5861824035644532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,4,1,power_law_1.2,0.03644160032272339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,4,1,power_law_1.01,3.5218177795410157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,4,1,power_law_1.2,0.044921600818634035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,4,1,power_law_1.2,0.06054400205612183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,4,1,power_law_1.2,0.07457919716835022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,4,1,power_law_1.2,0.08755840063095092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,4,1,power_law_1.2,0.14670720100402831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,4,1,power_law_1.2,0.1497215986251831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,4,1,power_law_1.2,0.15401599407196045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,4,1,power_law_1.2,0.15447039604187013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,4,1,power_law_1.2,0.15776000022888184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,4,1,power_law_1.2,0.17424000501632692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,4,1,power_law_1.2,0.18876800537109376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,4,1,power_law_1.2,0.19726719856262206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,4,1,power_law_1.2,0.19786880016326905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,4,1,power_law_1.2,0.02512640058994293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,4,1,power_law_1.2,0.21639039516448974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,4,1,power_law_1.2,0.2396224021911621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,4,1,power_law_1.2,0.31600639820098875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,4,1,power_law_1.2,0.3584896087646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,4,1,power_law_1.2,0.41692161560058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,4,1,power_law_1.2,0.5116991996765137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,4,1,power_law_1.2,0.6187583923339843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,4,1,power_law_1.2,0.028012800216674804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,4,1,power_law_1.2,0.0353408008813858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,4,1,power_law_1.2,0.04462080001831055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,4,1,power_law_1.2,0.05564799904823303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,4,1,power_law_1.2,0.07342079877853394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,4,1,power_law_1.2,0.07928959727287292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,4,1,power_law_1.2,0.8313535690307617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,4,1,power_law_1.2,0.8012928009033203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,4,1,power_law_1.2,1.3079615592956544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,4,1,power_law_1.2,0.08342400193214417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,4,1,power_law_1.2,0.08610560297966004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,4,1,power_law_1.2,0.08824959993362427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,4,1,power_law_1.2,0.09404159784317016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,4,1,power_law_1.2,0.10015360116958619
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,4,1,power_law_1.2,0.10503040552139283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,4,1,power_law_1.2,1.0461759567260742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,4,1,power_law_1.2,0.1176576018333435
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,4,1,power_law_1.2,0.10924160480499268
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,4,1,power_law_1.2,0.11437439918518066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,4,1,power_law_1.2,0.13056000471115112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,4,1,power_law_1.2,0.13322240114212036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,4,power_law_1.2,1.8797632217407227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,4,1,power_law_1.2,0.15448319911956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,4,1,power_law_1.2,0.16898560523986816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,4,1,power_law_1.2,0.19804799556732178
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,4,1,power_law_1.2,0.2863231897354126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,4,1,power_law_1.2,0.3081279993057251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,4,power_law_1.2,1.1228927612304687
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,4,1,power_law_1.2,0.4115456104278564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,4,balanced,0.060271998246510826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,4,balanced,0.0827946662902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,4,balanced,0.14684800306955972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,4,balanced,0.24908800919850668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,4,balanced,0.46025065581003827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,4,balanced,0.5818453232447306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,4,balanced,0.6117120186487833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,4,balanced,0.61299200852712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,4,1,power_law_1.2,1.4575615882873536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,4,balanced,0.6585706472396851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,4,balanced,0.6645386616388956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,4,balanced,0.654698650042216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,4,balanced,0.659877339998881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,4,balanced,0.6628693342208862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,4,balanced,0.6726079781850179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,4,balanced,0.681488037109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,4,balanced,0.6902506351470947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,4,balanced,0.7093706925710043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,4,balanced,0.835536003112793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,4,balanced,0.8675359884897867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,4,balanced,0.9370240370432535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,4,balanced,1.0130293369293213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,4,balanced,1.5414560635884602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,4,balanced,1.6969706217447917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,4,1,power_law_1.2,1.0172736167907714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,4,balanced,2.3916427294413247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,4,balanced,3.0464534759521484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,4,1,power_law_1.2,1.8675712585449218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,4,balanced,4.2521546681722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,4,balanced,5.816464106241862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,4,balanced,0.058506667613983154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,4,balanced,0.07843199868996938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,4,balanced,0.11211199561754863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,4,balanced,0.18095999956130981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,4,balanced,0.2926666736602783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,4,balanced,0.3593440055847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,4,balanced,0.4535626570383708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,4,balanced,0.41019733746846515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,4,balanced,0.3808319965998332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,4,balanced,0.3839786847432454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,4,balanced,0.38766932487487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,4,balanced,0.3960053523381551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,4,balanced,0.40024534861246747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,4,balanced,0.41284799575805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,4,balanced,0.4241120020548503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,4,balanced,0.4351573387781779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,4,1,power_law_1.2,2.692403221130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,4,balanced,0.4569013516108195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,4,balanced,0.5363466739654541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,4,balanced,0.5741653442382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,4,balanced,0.6469120184580485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,4,balanced,0.7189173698425293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,4,balanced,1.0557759602864583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,4,balanced,1.2044800122578938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,4,balanced,1.7053920427958171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,4,balanced,2.1751413345336914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,4,balanced,2.9861119588216147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,4,1,power_law_1.2,1.8420223236083983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,4,balanced,3.9950027465820312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,4,1,power_law_1.2,0.5280704021453857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,4,power_law_1.2,5.0373695373535154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,4,power_law_1.2,3.1349504470825194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,4,power_law_1.01,0.14469759464263915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,4,balanced,0.03121600051720937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,4,balanced,0.04137066751718521
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,4,balanced,0.06435200075308482
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,4,balanced,0.0925386647383372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,4,balanced,0.15172800421714783
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,4,balanced,0.18093866109848022
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,4,balanced,0.18505066633224487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,4,balanced,0.1892319917678833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,4,balanced,0.19330666462580362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,4,balanced,0.18921067317326865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,4,balanced,0.19180800517400107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,4,balanced,0.19547200202941895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,4,balanced,0.19260799884796143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,4,balanced,0.19946666558583578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,4,balanced,0.20574400822321573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,4,balanced,0.25549866755803424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,4,balanced,0.23607999086380005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,4,balanced,0.5044639905293783
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,4,balanced,0.3768693208694458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,4,power_law_1.01,0.23056640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,4,balanced,0.2606186668078105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,4,balanced,0.28005866209665936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,4,power_law_1.01,0.26089599132537844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,4,balanced,0.4103253285090129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,4,power_law_1.01,0.3163136005401611
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,4,balanced,0.4416053295135498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,4,power_law_1.01,0.34483840465545657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,4,power_law_1.01,0.525817584991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,4,power_law_1.01,0.5356927871704101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,4,balanced,0.06450133522351582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,4,power_law_1.01,0.5500288009643555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,4,power_law_1.01,0.6216320037841797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,4,power_law_1.01,0.6343616008758545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,4,balanced,0.6272053321202596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,4,power_law_1.01,0.6102272033691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,4,balanced,0.8110880057017008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,4,power_law_1.01,0.627129602432251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,4,balanced,1.0847626527150471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,4,power_law_1.01,0.6438079833984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,4,balanced,1.6598985989888508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,4,power_law_1.01,0.6572991847991944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,4,1,power_law_1.2,1.4546943664550782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,4,power_law_1.01,0.6606592178344727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,4,power_law_1.01,0.0913536012172699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,4,power_law_1.01,0.6870528221130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,4,power_law_1.01,0.7375296115875244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,4,power_law_1.01,0.8557632446289063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,4,power_law_1.01,0.9482751846313476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,4,power_law_1.01,1.133580780029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,4,power_law_1.01,0.1598080039024353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,4,power_law_1.01,0.1609663963317871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,4,power_law_1.01,1.2872896194458008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,4,power_law_1.01,0.19568639993667603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,4,power_law_1.01,0.234335994720459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,4,power_law_1.01,1.6327104568481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,4,power_law_1.01,1.9821760177612304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,4,power_law_1.01,2.7867136001586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,4,power_law_1.01,0.28758399486541747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,4,power_law_1.01,0.32782080173492434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,4,power_law_1.01,0.315449595451355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,4,power_law_1.01,0.3351680040359497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,4,power_law_1.01,0.3486079931259155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,4,power_law_1.01,3.603084945678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,4,power_law_1.01,0.35999360084533694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,4,power_law_1.01,0.3760960102081299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,4,power_law_1.01,0.38487040996551514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,4,power_law_1.01,0.38991360664367675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,4,balanced,0.09795733292897542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,4,balanced,0.10454400380452473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,4,balanced,0.17506666978200278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,4,power_law_1.01,5.073862457275391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,4,power_law_1.01,0.4127808094024658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,4,1,power_law_1.2,1.889414405822754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,4,balanced,0.3064639965693156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,4,power_law_1.01,0.4247231960296631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,4,balanced,0.35542933146158856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,4,balanced,0.3596746524175008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,4,power_law_1.01,0.46606078147888186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,4,power_law_1.01,6.930540466308594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,4,balanced,0.3641599814097087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,4,balanced,0.36988266309102374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,4,power_law_1.01,0.5553088188171387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,4,power_law_1.01,0.6205440044403077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,4,power_law_1.01,0.7406847953796387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,4,power_law_1.01,0.8689536094665528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,4,power_law_1.01,1.0844479560852052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,4,power_law_1.01,1.3313983917236327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,4,power_law_1.01,1.8700607299804688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,4,1,power_law_1.2,0.730624008178711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,4,power_law_1.01,2.3891071319580077
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,4,power_law_1.01,0.057811200618743896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,4,power_law_1.01,3.3155391693115233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,4,power_law_1.01,0.08481919765472412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,4,power_law_1.01,0.08928639888763427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,4,power_law_1.01,0.10247039794921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,4,power_law_1.01,0.12725119590759276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,4,power_law_1.01,0.16165759563446044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,4,power_law_1.01,4.4336193084716795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,4,power_law_1.01,0.15820800065994262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,4,power_law_1.01,0.16993279457092286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,4,power_law_1.2,0.1467519998550415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,4,power_law_1.01,0.17455359697341918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,4,power_law_1.2,0.17455999851226806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,4,power_law_1.01,0.1663807988166809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,4,power_law_1.2,0.27070720195770265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,4,power_law_1.01,0.18026880025863648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,4,power_law_1.2,0.3179840087890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,4,power_law_1.01,0.18154879808425903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,4,power_law_1.01,0.17779200077056884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,4,power_law_1.2,0.34656639099121095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,4,power_law_1.01,0.19059840440750123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,4,power_law_1.2,0.5150271892547608
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,4,power_law_1.01,0.20177919864654542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,4,power_law_1.2,0.5101503849029541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,4,power_law_1.01,0.20811519622802735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,4,power_law_1.01,0.21635839939117432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,4,power_law_1.01,0.28072960376739503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,4,power_law_1.01,0.3026432037353516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,4,power_law_1.01,0.3389440059661865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,4,power_law_1.01,0.378656005859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,4,power_law_1.01,0.42919039726257324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,4,power_law_1.2,0.5021312236785889
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,4,power_law_1.01,0.5643775939941407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,4,balanced,0.4177546501159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,4,power_law_1.2,0.6295872211456299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,4,balanced,0.4254773457845052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,4,power_law_1.01,0.7482944011688233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,4,balanced,0.4334133466084798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,4,power_law_1.2,0.640064001083374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,4,balanced,0.43750933806101483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,4,power_law_1.2,0.6194752216339111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,4,power_law_1.01,0.9796544075012207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,4,power_law_1.2,0.6466176033020019
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,4,power_law_1.01,1.2867391586303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,4,power_law_1.2,0.6597695827484131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,4,power_law_1.2,0.6821184158325195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,4,power_law_1.01,2.0748287200927735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,4,power_law_1.2,0.6975615978240967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,4,power_law_1.2,0.6965439796447754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,4,power_law_1.2,0.09221760034561158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,4,power_law_1.2,0.13545600175857545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,4,power_law_1.2,0.7667391777038575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,4,power_law_1.2,0.15820800065994262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,4,1,power_law_1.2,3.5203712463378904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,4,power_law_1.2,0.9040384292602539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,4,power_law_1.2,0.2035968065261841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,4,power_law_1.2,0.22641279697418212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,4,power_law_1.01,0.10101120471954346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,4,power_law_1.2,0.9819583892822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,4,power_law_1.2,0.2765952110290527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,4,power_law_1.2,0.32522881031036377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,4,power_law_1.2,1.1700927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,4,power_law_1.2,0.32969601154327394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,4,power_law_1.2,0.35587201118469236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,4,power_law_1.2,1.356223964691162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,4,power_law_1.2,0.34524800777435305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,4,power_law_1.2,0.3772351980209351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,4,1,power_law_1.2,2.3829376220703127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,4,power_law_1.2,1.6946048736572266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,4,power_law_1.2,2.0588735580444335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,4,power_law_1.2,0.3814912080764771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,4,power_law_1.2,0.3903359889984131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,4,power_law_1.2,0.4135104179382324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,4,power_law_1.2,0.41786880493164064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,4,power_law_1.2,0.4328256130218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,4,1,power_law_1.2,1.0017727851867675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,4,1,balanced,0.05604266623655955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,4,power_law_1.2,0.48511362075805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,4,1,balanced,0.08358933528264363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,4,1,balanced,0.14442666371663412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,4,1,balanced,0.2534880042076111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,4,power_law_1.2,0.5815296173095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,4,balanced,0.6250186761220297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,4,balanced,0.639578660329183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,4,power_law_1.2,0.6399295806884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,4,balanced,0.6538613239924113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,4,balanced,0.6932586828867594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,4,power_law_1.2,0.7506048202514648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,4,balanced,0.7459519704182943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,4,balanced,0.80294402440389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,4,power_law_1.2,0.9085824012756347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,4,power_law_1.2,0.05867519974708557
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,4,power_law_1.2,0.06154239773750305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,4,power_law_1.2,1.1187199592590331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,4,1,balanced,0.4737866719563802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,4,1,balanced,0.5964959859848022
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,4,power_law_1.2,0.08477439880371093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,4,1,balanced,0.6236159801483154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,4,power_law_1.2,0.09946879744529724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,4,1,balanced,0.6281599998474121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,4,power_law_1.2,0.10764800310134888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,4,power_law_1.2,0.15939199924468994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,4,power_law_1.2,0.1637503981590271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,4,power_law_1.2,2.9442623138427733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,4,power_law_1.2,0.16595200300216675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,4,power_law_1.2,0.17883520126342772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,4,power_law_1.2,0.1732800006866455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,4,power_law_1.2,0.18654719591140748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,4,power_law_1.2,1.3965439796447754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,4,power_law_1.2,0.1895616054534912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,4,power_law_1.01,0.18954240083694457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,4,power_law_1.2,0.18417279720306395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,4,power_law_1.01,0.16383999586105347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,4,power_law_1.2,0.10062719583511352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,4,power_law_1.2,0.19871360063552856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,4,power_law_1.01,0.21477758884429932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,4,power_law_1.2,2.0378368377685545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,4,power_law_1.2,0.20723841190338135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,4,power_law_1.01,0.22294399738311768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,4,power_law_1.2,0.2352128028869629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,4,power_law_1.01,0.3347264051437378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,4,power_law_1.2,0.22699520587921143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,4,power_law_1.01,0.3251904010772705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,4,power_law_1.2,0.3525824069976807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,4,power_law_1.01,0.3417407989501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,4,1,balanced,0.6325600147247314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,4,1,balanced,0.6346346537272135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,4,power_law_1.01,0.3477440118789673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,4,1,balanced,0.6393653154373169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,4,1,balanced,0.6377386649449667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,4,1,balanced,0.6403839985529581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,4,1,balanced,0.6430879831314087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,4,1,balanced,0.6516266663869222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,4,1,balanced,0.6595093409220377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,4,1,balanced,0.6731893221537272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,4,1,balanced,0.8166879812876383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,4,power_law_1.2,0.36050560474395754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,4,1,balanced,0.8395626544952393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,4,power_law_1.2,0.34897921085357664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,4,1,balanced,0.8872799873352051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,4,power_law_1.2,0.3776895999908447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,4,1,balanced,0.9444693724314371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,4,power_law_1.2,2.336345672607422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,4,power_law_1.2,0.5201727867126464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,4,1,balanced,1.4104852676391602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,4,power_law_1.2,0.5676032066345215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,4,1,balanced,1.5175946553548176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,4,power_law_1.2,0.7621568202972412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,4,power_law_1.2,4.0034431457519535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,4,1,balanced,2.2354933420817056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,4,power_law_1.2,2.380966377258301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,4,power_law_1.2,1.0790271759033203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,4,1,balanced,0.04455466568470001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,4,1,balanced,0.06032533446947733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,4,1,balanced,0.08744000395139058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,4,1,balanced,0.16682666540145874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,4,1,balanced,2.8584105173746743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,4,1,balanced,0.27900266647338867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,4,1,balanced,0.34436798095703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,4,1,balanced,0.39950935045878094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,4,1,balanced,0.3994506597518921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,4,1,balanced,0.3727946678797404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,4,1,balanced,0.3734293381373088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,4,1,balanced,3.632042566935221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,4,power_law_1.2,5.1652992248535154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,4,1,balanced,5.046079953511556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,4,1,balanced,0.3694560130437215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,4,1,balanced,0.37142932415008545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,4,1,balanced,0.3752640088399251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,4,1,balanced,0.3797599871953328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,4,1,balanced,0.0558186670144399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,4,1,balanced,0.0745066652695338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,4,1,balanced,0.3878026803334554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,4,1,balanced,0.10357866684595744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,4,1,balanced,0.3935626745223999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,4,1,balanced,0.18615466356277466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,4,power_law_1.2,0.1257024049758911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,4,1,balanced,0.30769066015879315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,4,power_law_1.2,6.8337150573730465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,4,power_law_1.2,0.18824319839477538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,4,power_law_1.2,0.20992000102996827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,4,power_law_1.2,0.22529280185699463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,4,power_law_1.2,0.3039423942565918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,4,power_law_1.2,0.330348801612854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,4,power_law_1.2,0.3362368106842041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,4,1,balanced,0.4058719873428345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,4,power_law_1.2,0.3533375978469849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,4,1,balanced,0.3764959971110026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,4,1,balanced,0.5169013341267904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,4,1,balanced,0.3574506839116414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,4,1,balanced,0.5373226801554362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,4,1,balanced,0.3575199842453003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,4,1,balanced,0.5827039877573649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,4,1,balanced,0.35846932729085285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,4,1,balanced,0.3599199851353963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,4,1,balanced,0.6287093162536621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,4,1,balanced,0.36287999153137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,4,1,balanced,0.3663253386815389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,4,1,balanced,0.9519039789835612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,4,1,balanced,0.36897599697113037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,4,1,balanced,0.3765546480814616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,4,1,balanced,1.0492053031921387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,4,1,balanced,0.3853013515472412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,4,1,balanced,0.5457546710968018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,4,1,balanced,1.4950613975524902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,4,power_law_1.01,0.3934528112411499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,4,balanced,1.2777067025502522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,4,power_law_1.01,0.41571841239929197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,4,power_law_1.01,0.4140480041503906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,4,balanced,1.3822933832804363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,4,1,balanced,0.03161599983771642
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,4,1,balanced,0.04311466713746389
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,4,1,balanced,0.0634933312733968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,4,power_law_1.01,0.4537407875061035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,4,1,balanced,0.09480533003807068
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,4,1,balanced,0.15478400389353433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,4,1,balanced,0.18796267112096152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,4,balanced,1.987178643544515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,4,power_law_1.01,0.6009151935577393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,4,power_law_1.01,0.6689023971557617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,4,balanced,2.529967943827311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,4,power_law_1.2,3.487776184082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,4,power_law_1.01,0.681990385055542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,4,1,balanced,1.90937074025472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,4,power_law_1.01,0.7606912136077881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,4,1,balanced,2.514773368835449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,4,balanced,3.348378817240397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,4,power_law_1.01,0.8828160285949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,4,1,balanced,0.18995199600855509
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,4,1,balanced,0.1974453330039978
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,4,1,balanced,0.2000746726989746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,4,1,balanced,0.20363199710845947
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,4,1,balanced,0.20619734128316244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,4,1,balanced,3.382746696472168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,4,1,balanced,0.21063466866811117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,4,1,balanced,0.2514080007870992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,4,balanced,4.4093014399210615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,4,1,balanced,0.26021866003672284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,4,1,balanced,0.26875199874242145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,4,1,balanced,0.26090667645136517
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,4,1,balanced,0.2770880063374837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,4,1,balanced,0.29705599943796795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,4,1,balanced,0.31669867038726807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,4,1,balanced,0.45586133003234863
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,4,1,balanced,0.41196266810099286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,4,1,balanced,0.9225333531697592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,4,balanced,6.315055847167969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,4,1,balanced,0.6799786885579427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,4,1,balanced,0.980778694152832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,4,1,power_law_1.01,0.05651199817657471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,4,1,power_law_1.01,0.08140799999237061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,4,1,balanced,1.2876799901326497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,4,1,power_law_1.01,0.1441215991973877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,4,1,balanced,1.2762666543324788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,4,1,power_law_1.01,0.22442240715026857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,4,power_law_1.2,1.4241215705871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,4,1,power_law_1.01,0.299782395362854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,4,1,balanced,1.8443466822306316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,4,1,power_law_1.01,0.4596735954284668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,4,balanced,8.26369603474935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,4,1,power_law_1.01,0.04336639940738678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,4,1,power_law_1.01,0.5290239810943603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,4,1,power_law_1.01,0.059935998916625974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,4,1,power_law_1.01,0.5350783824920654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,4,1,power_law_1.01,0.08880640268325805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,4,1,power_law_1.01,0.1344063997268677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,4,1,power_law_1.01,0.18103040456771852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,4,1,power_law_1.01,0.05242239832878113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,4,1,power_law_1.01,0.27723519802093505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,4,1,power_law_1.01,0.07200000286102295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,4,1,power_law_1.01,0.09873279929161072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,4,power_law_1.2,0.39530239105224607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,4,power_law_1.2,0.42674560546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,4,power_law_1.2,0.41521282196044923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,4,1,power_law_1.01,0.5835072040557862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,4,power_law_1.2,0.47543678283691404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,4,1,power_law_1.01,0.590015983581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,4,power_law_1.2,0.6428544044494628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,4,1,power_law_1.01,0.5915711879730224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,4,1,power_law_1.01,0.33803520202636717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,4,1,power_law_1.01,0.6035711765289307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,4,power_law_1.2,0.6711167812347412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,4,1,power_law_1.01,0.14106240272521972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,4,1,power_law_1.01,0.35396480560302734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,4,1,power_law_1.01,0.6034495830535889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,4,1,power_law_1.01,0.21001598834991456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,4,power_law_1.2,0.6861440181732178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,4,1,power_law_1.01,0.30056960582733155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,4,1,power_law_1.01,0.6154176235198975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,4,power_law_1.2,0.8027008056640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,4,1,power_law_1.01,0.32372479438781737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,4,1,power_law_1.01,0.6252543926239014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,4,1,power_law_1.01,0.33033599853515627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,4,power_law_1.2,0.9666751861572266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,4,1,power_law_1.01,0.6339136123657226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,4,1,balanced,0.4879466692606608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,4,1,power_law_1.01,0.3292351961135864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,4,1,balanced,0.856229305267334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,4,1,power_law_1.01,0.3305408000946045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,4,power_law_1.2,1.048256015777588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,4,1,balanced,0.7293919722239176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,4,1,power_law_1.01,0.34921600818634035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,4,1,power_law_1.01,0.3440576076507568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,4,1,power_law_1.01,0.36329600811004636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,4,1,power_law_1.01,0.34452478885650634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,4,1,power_law_1.01,0.39107840061187743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,4,1,power_law_1.01,0.346943998336792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,4,1,power_law_1.01,0.42142720222473146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,4,1,power_law_1.01,0.3519167900085449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,4,1,power_law_1.01,0.4634431838989258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,4,1,power_law_1.01,0.35895040035247805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,4,1,power_law_1.01,0.5010496139526367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,4,1,power_law_1.01,0.36828160285949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,4,1,power_law_1.01,0.3751807928085327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,4,1,balanced,1.1599679787953694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,4,1,power_law_1.01,0.3845887899398804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,4,1,power_law_1.01,0.41439361572265626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,4,1,balanced,1.2251520156860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,4,power_law_1.2,2.3777791976928713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,4,1,power_law_1.01,0.5204800128936767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,4,1,balanced,1.7815146446228027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,4,1,power_law_1.01,0.5761919975280761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,4,1,power_law_1.01,0.6850880146026611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,4,1,power_law_1.01,0.6631423950195312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,4,1,balanced,2.331221262613932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,4,1,power_law_1.01,0.8239423751831054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,4,1,power_law_1.01,0.7533184051513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,4,1,power_law_1.01,0.8921536445617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,4,1,balanced,2.966815948486328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,4,1,power_law_1.01,0.9396224021911621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,4,1,power_law_1.01,1.013043212890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,4,1,power_law_1.01,1.1226367950439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,4,1,power_law_1.01,1.1422719955444336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,4,1,balanced,3.9891945521036782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,4,1,power_law_1.01,1.4909695625305175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,4,1,power_law_1.01,1.3911999702453612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,4,power_law_1.2,4.492275238037109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,4,1,power_law_1.01,1.8834112167358399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,4,1,power_law_1.2,0.05454080104827881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,4,1,balanced,5.645994822184245
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,4,1,power_law_1.01,0.02784000039100647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,4,1,power_law_1.2,0.08319360017776489
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,4,1,power_law_1.01,0.041529598832130435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,4,1,power_law_1.2,0.14615039825439452
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,4,1,power_law_1.01,0.06135680079460144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,4,1,power_law_1.2,0.20509440898895265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,4,1,power_law_1.01,0.07998719811439514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,4,1,power_law_1.01,2.6345535278320313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,4,1,power_law_1.2,0.258790397644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,4,power_law_1.01,1.0967231750488282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,4,1,power_law_1.01,0.10631680488586426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,4,1,power_law_1.2,0.4624767780303955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,4,1,power_law_1.01,0.15432319641113282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,4,1,power_law_1.01,0.16791679859161376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,4,1,power_law_1.2,0.5089791774749756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,4,1,power_law_1.01,1.6605440139770509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,4,1,balanced,7.241269429524739
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,4,1,power_law_1.01,0.174399995803833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,4,1,power_law_1.2,0.5298943996429444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,4,1,power_law_1.01,3.406355285644531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,4,1,power_law_1.01,0.1822335958480835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,4,1,power_law_1.2,0.041126400232315063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,4,1,power_law_1.2,0.5790143966674804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,4,1,power_law_1.01,0.18613760471343993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,4,1,power_law_1.2,0.05914880037307739
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,4,1,power_law_1.01,0.19882240295410156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,4,1,power_law_1.01,2.177952003479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,4,1,power_law_1.2,0.5865407943725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,4,1,power_law_1.2,0.08727679848670959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,4,1,power_law_1.01,0.2129983901977539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,4,1,power_law_1.2,0.12723840475082399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,4,1,power_law_1.2,0.6028160095214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,4,1,power_law_1.01,0.22740480899810792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,4,1,power_law_1.2,0.17002240419387818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,4,1,power_law_1.2,0.054604798555374146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,4,1,power_law_1.01,0.26017920970916747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,4,1,power_law_1.2,0.6059391975402832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,4,1,power_law_1.2,0.2731584072113037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,4,1,power_law_1.2,0.07255679965019227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,4,1,power_law_1.01,0.2896575927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,4,1,power_law_1.01,2.8166528701782227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,4,1,power_law_1.2,0.6146240234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,4,1,power_law_1.2,0.34235520362854005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,4,1,power_law_1.2,0.0972544014453888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,4,1,power_law_1.2,0.3428096055984497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,4,1,power_law_1.2,0.6244607925415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,4,1,power_law_1.2,0.13040000200271606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,4,1,power_law_1.2,0.3514431953430176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,4,1,power_law_1.2,0.20164480209350585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,4,1,power_law_1.2,0.6300735950469971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,4,1,power_law_1.2,0.3530751943588257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,4,1,power_law_1.2,0.30099198818206785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,4,1,power_law_1.2,0.6588992118835449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,4,1,power_law_1.2,0.3500159978866577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,4,1,power_law_1.01,3.8620033264160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,4,1,power_law_1.2,0.32483201026916503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,4,1,power_law_1.2,0.3526655912399292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,4,1,power_law_1.2,0.7014976024627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,4,1,power_law_1.2,0.32799999713897704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,4,1,power_law_1.2,0.35910398960113527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,4,1,power_law_1.2,0.3323712110519409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,4,1,power_law_1.2,0.8655872344970703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,4,1,power_law_1.2,0.370796799659729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,4,1,power_law_1.2,0.34012799263000487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,4,1,power_law_1.2,0.3762687921524048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,4,1,power_law_1.2,0.9167039871215821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,4,1,power_law_1.2,0.3559231996536255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,4,1,power_law_1.2,0.3924992084503174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,4,1,power_law_1.2,0.3861952066421509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,4,1,power_law_1.2,1.0385408401489258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,4,1,power_law_1.2,0.42880001068115237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,4,1,power_law_1.01,5.114553451538086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,4,1,power_law_1.2,0.39555840492248534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,4,1,power_law_1.2,0.5439295768737793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,4,1,power_law_1.2,0.580294418334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,4,1,power_law_1.2,0.673747205734253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,4,1,power_law_1.2,1.1594880104064942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,4,1,power_law_1.2,0.7655104160308838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,4,1,power_law_1.01,0.6360960006713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,4,1,power_law_1.01,0.24527359008789062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,4,1,power_law_1.2,1.4187456130981446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,4,1,power_law_1.01,0.2785856008529663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,4,1,power_law_1.01,0.7935679912567138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,4,1,power_law_1.2,0.950432014465332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,4,power_law_1.2,1.3969663619995116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,4,1,power_law_1.01,0.30053761005401614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,4,1,power_law_1.01,0.3291584014892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,4,1,power_law_1.01,0.9211520195007324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,4,1,power_law_1.2,1.140659236907959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,4,1,power_law_1.2,1.6805952072143555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,4,power_law_1.2,1.706675148010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,4,1,power_law_1.01,0.3865855932235718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,4,1,power_law_1.2,0.4385663986206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,4,1,power_law_1.01,1.1317248344421387
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,4,1,power_law_1.01,0.4536320209503174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,4,1,power_law_1.2,1.5113408088684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,4,1,power_law_1.2,0.47499518394470214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,4,1,power_law_1.01,0.45584640502929685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,4,1,power_law_1.2,0.031052801012992858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,4,1,power_law_1.2,2.215033531188965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,4,1,power_law_1.2,0.5158656120300293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,4,1,power_law_1.01,1.3482111930847167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,4,power_law_1.2,2.408095932006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,4,1,power_law_1.2,0.04565759897232056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,4,1,power_law_1.01,0.549894380569458
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,4,1,power_law_1.2,0.06170880198478699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,4,1,power_law_1.01,0.74236159324646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,4,1,power_law_1.2,0.07806079983711242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,4,1,power_law_1.2,0.10316799879074097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,4,power_law_1.2,3.0211008071899412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,4,1,power_law_1.2,0.15637760162353515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,4,1,power_law_1.2,0.16647039651870726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,4,1,power_law_1.2,2.732627105712891
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,4,1,power_law_1.2,0.18105599880218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,4,1,power_law_1.2,1.896678352355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,4,1,power_law_1.2,0.1861248016357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,4,1,power_law_1.2,0.1945472002029419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,4,power_law_1.2,4.041350555419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,4,1,power_law_1.2,0.20728321075439454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,4,1,power_law_1.01,1.7748096466064454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,4,1,power_law_1.2,0.22272639274597167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,4,1,power_law_1.2,2.6547008514404298
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,4,1,power_law_1.01,0.9429120063781739
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,4,1,power_law_1.2,0.2364351987838745
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,4,1,power_law_1.2,0.2676160097122192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,4,1,power_law_1.01,2.1918272018432616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,4,1,power_law_1.01,1.3168704032897949
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,4,1,power_law_1.2,0.29999361038208006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,4,1,power_law_1.2,3.9255489349365233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,4,1,power_law_1.2,0.2540607929229736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,4,1,power_law_1.2,0.28862719535827636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,4,1,power_law_1.01,1.8434368133544923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,4,1,power_law_1.01,3.0196672439575196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,4,1,power_law_1.2,0.2980288028717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,4,balanced,0.06562133133411407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,4,balanced,0.09081600109736125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,4,1,power_law_1.2,0.3438976049423218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,4,balanced,0.15573333700497946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,4,balanced,0.2488480011622111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,4,balanced,0.4371680021286011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,4,1,power_law_1.2,0.36884479522705077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,4,balanced,0.8462879657745361
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,4,1,power_law_1.2,0.4267712116241455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,4,balanced,1.2851413091023762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,4,1,power_law_1.01,3.8544254302978516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,4,1,power_law_1.2,0.46460800170898436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,4,balanced,1.2864373524983723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,4,1,power_law_1.2,0.5691904067993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,4,balanced,1.4149813652038574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,4,balanced,1.421493371327718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,4,power_law_1.01,1.3371264457702636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,4,balanced,1.4399199485778809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,4,power_law_1.01,1.6356800079345704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,4,balanced,1.4513866106669109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,4,1,power_law_1.01,5.501299285888672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,4,balanced,1.4559359550476074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,4,1,power_law_1.2,0.7606783866882324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,4,power_law_1.2,5.161433410644531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,4,1,power_law_1.2,3.4167808532714843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,4,balanced,1.455466588338216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,4,power_law_1.01,2.34704647064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,4,1,power_law_1.2,0.9538304328918457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,4,balanced,0.053786665201187134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,4,balanced,1.4719732602437336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,4,balanced,0.068122665087382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,4,balanced,0.09498133261998494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,4,balanced,0.15244799852371216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,4,balanced,0.27076266209284466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,4,balanced,1.483839988708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,4,balanced,0.49196799596150714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,4,power_law_1.01,2.7921728134155273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,4,balanced,0.8016746838887533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,4,balanced,1.504053274790446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,4,balanced,0.8030347029368082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,4,balanced,0.7689279715220133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,4,1,power_law_1.01,7.165261077880859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,4,balanced,1.7326773007710774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,4,balanced,0.7729600270589193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,4,balanced,0.8126400311787924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,4,balanced,1.7652373313903809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,4,balanced,0.8282506465911865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,4,power_law_1.01,4.144908905029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,4,balanced,0.8367893695831299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,4,balanced,1.835312048594157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,4,balanced,0.8480587005615234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,4,balanced,0.8616586526234945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,4,balanced,1.9069600105285645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,4,balanced,0.8760533332824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,4,balanced,0.07242133220036824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,4,balanced,0.9002453486124674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,4,balanced,0.08612799644470215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,4,1,power_law_1.2,0.6542912006378174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,4,balanced,0.1071573297182719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,4,balanced,0.1591146687666575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,4,balanced,1.050597349802653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,4,balanced,0.2760319908459981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,4,1,power_law_1.2,0.8138431549072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,4,balanced,0.5091893275578817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,4,power_law_1.01,5.354105758666992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,4,balanced,1.082858641942342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,4,balanced,0.7374080022176107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,4,balanced,0.7402506669362386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,4,1,power_law_1.2,0.9387136459350586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,4,balanced,1.2452479998270671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,4,1,power_law_1.2,1.1511232376098632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,4,balanced,1.2511093616485596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,4,balanced,2.0487146377563477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,4,balanced,1.260970671971639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,4,1,power_law_1.2,1.359001636505127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,4,balanced,1.270645300547282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,4,balanced,2.1921119689941406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,4,balanced,1.281392018000285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,4,1,power_law_1.2,1.7843584060668944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,4,balanced,1.153066635131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,4,power_law_1.01,7.446041870117187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,4,balanced,1.2964800198872883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,4,balanced,2.4933172861735025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,4,balanced,1.316912015279134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,4,1,power_law_1.2,2.2050432205200194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,4,balanced,1.3335413932800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,4,balanced,3.621253331502279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,4,balanced,1.366960048675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,4,balanced,1.4243359565734863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,4,1,power_law_1.2,3.0297216415405273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,4,balanced,1.482634703318278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,4,balanced,1.227461338043213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,4,balanced,4.291712125142415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,4,balanced,1.5899839401245117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,4,balanced,1.3689066569010417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,4,power_law_1.01,9.471878051757812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,4,balanced,1.5140533447265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,4,balanced,1.701050599416097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,4,1,power_law_1.2,3.842559814453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,4,balanced,1.8068532943725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,4,balanced,1.92193603515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,4,balanced,5.927279790242513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,4,power_law_1.2,7.807430267333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,4,balanced,2.5088586807250977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,4,power_law_1.01,0.1581120014190674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,4,balanced,2.96946652730306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,4,power_law_1.01,0.24960639476776122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,4,balanced,3.140495936075846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,4,power_law_1.01,0.43144960403442384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,4,balanced,3.3971147537231445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,4,power_law_1.01,0.4804671764373779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,4,power_law_1.01,0.6133952140808105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,4,balanced,4.221904118855794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,4,balanced,4.696821212768555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,4,power_law_1.01,0.8213952064514161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,4,power_law_1.01,0.08895360231399536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,4,1,power_law_1.2,4.914438247680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,4,power_law_1.01,0.15189759731292723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,4,power_law_1.01,0.9386112213134765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,4,power_law_1.01,0.26264960765838624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,4,power_law_1.01,0.26557440757751466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,4,1,power_law_1.2,5.507263946533203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,4,power_law_1.01,1.0482175827026368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,4,balanced,6.405866622924805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,4,power_law_1.01,0.3733952045440674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,4,1,power_law_1.2,1.3349311828613282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,4,power_law_1.01,1.2028223991394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,4,power_law_1.01,0.417632007598877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,4,power_law_1.01,0.09934719800949096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,4,power_law_1.01,0.4976128101348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,4,power_law_1.01,1.194694423675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,4,power_law_1.01,0.1565567970275879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,4,power_law_1.01,0.663097620010376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,4,balanced,0.0315733328461647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,4,balanced,0.03980266551176707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,4,power_law_1.01,0.2666368007659912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,4,power_law_1.01,1.2059136390686036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,4,balanced,0.062128002444903054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,4,balanced,0.09277333815892537
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,4,balanced,0.1502240002155304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,4,power_law_1.01,0.6117311954498291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,4,balanced,0.25618666410446167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,4,balanced,0.3632853428522746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,4,power_law_1.01,1.1276927947998048
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,4,balanced,0.3667626778284709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,4,balanced,8.965775807698568
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,4,balanced,0.369701345761617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,4,power_law_1.01,0.6843967914581299
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,4,balanced,0.3739039897918701
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,4,balanced,0.379258672396342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,4,power_law_1.01,1.2029760360717774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,4,balanced,0.3861493269602458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,4,power_law_1.01,0.7357312202453613
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,4,balanced,0.3957600196202596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,4,balanced,0.38899731636047363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,4,balanced,0.3970186710357666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,4,power_law_1.01,1.2062463760375977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,4,power_law_1.01,0.6958784103393555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,4,balanced,0.4063200155893962
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,4,balanced,0.406335989634196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,4,balanced,0.42868268489837646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,4,power_law_1.01,0.2687936067581177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,4,power_law_1.01,0.7194943904876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,4,power_law_1.01,1.2208959579467773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,4,power_law_1.01,0.37764480113983157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,4,power_law_1.01,0.7485119819641113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,4,power_law_1.01,1.2585023880004882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,4,power_law_1.01,0.49377918243408203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,4,power_law_1.01,0.7233535766601562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,4,power_law_1.01,0.4982912063598633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,4,power_law_1.01,0.805907154083252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,4,power_law_1.01,0.6459136009216309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,4,power_law_1.01,0.8064000129699707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,4,balanced,0.4971199830373128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,4,balanced,0.4712800184885661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,4,power_law_1.01,1.0523712158203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,4,balanced,0.9098400274912516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,4,balanced,0.5794133345286051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,4,balanced,0.5875786542892456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,4,power_law_1.01,1.045248031616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,4,power_law_1.01,1.2918208122253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,4,power_law_1.01,1.0974464416503906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,4,power_law_1.01,1.5203136444091796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,4,power_law_1.01,1.0710016250610352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,4,power_law_1.01,0.9410495758056641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,4,power_law_1.01,1.5318143844604493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,4,power_law_1.01,1.1022784233093261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,4,power_law_1.01,0.9974911689758301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,4,balanced,0.6849706967671713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,4,power_law_1.01,0.05616639852523804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,4,power_law_1.01,1.7184064865112305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,4,power_law_1.01,1.1354368209838868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,4,power_law_1.01,1.1351167678833007
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,4,balanced,0.9652106761932373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,4,power_law_1.01,0.08357759714126586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,4,balanced,1.1466986338297527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,4,power_law_1.01,1.2313088417053222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,4,power_law_1.01,0.13841279745101928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,4,power_law_1.01,1.9219327926635743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,4,power_law_1.01,0.14507520198822021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,4,power_law_1.01,1.4785728454589844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,4,power_law_1.01,0.2120448112487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,4,1,power_law_1.2,1.9274560928344726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,4,power_law_1.01,2.2556095123291016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,4,power_law_1.01,0.2254080057144165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,4,power_law_1.01,1.7204544067382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,4,power_law_1.2,0.15285760164260864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,4,power_law_1.01,0.24987521171569824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,4,power_law_1.2,0.24744319915771484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,4,power_law_1.01,1.1380607604980468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,4,power_law_1.01,2.596063995361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,4,power_law_1.01,0.3220479965209961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,4,power_law_1.01,2.1600704193115234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,4,power_law_1.2,0.36343040466308596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,4,balanced,1.7992053031921387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,4,power_law_1.01,1.2042495727539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,4,power_law_1.01,0.2982912063598633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,4,power_law_1.2,0.4217855930328369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,4,power_law_1.2,11.213728332519532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,4,power_law_1.01,0.3034879922866821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,4,power_law_1.2,0.0877888023853302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,4,power_law_1.01,1.2215935707092285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,4,power_law_1.2,0.6071807861328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,4,power_law_1.01,2.6339839935302733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,4,power_law_1.01,3.255411148071289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,4,power_law_1.2,0.1502527952194214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,4,power_law_1.01,0.2888000011444092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,4,power_law_1.2,0.21111040115356444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,4,power_law_1.01,1.3698880195617675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,4,power_law_1.01,0.32936959266662597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,4,power_law_1.2,0.7142079830169678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,4,power_law_1.2,0.26664319038391116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,4,power_law_1.01,0.36992640495300294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,4,power_law_1.2,0.8317248344421386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,4,power_law_1.01,1.5244352340698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,4,power_law_1.01,3.5414783477783205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,4,power_law_1.2,0.3638592004776001
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,4,power_law_1.01,0.32120959758758544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,4,power_law_1.01,3.890943908691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,4,power_law_1.2,1.0864768028259277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,4,power_law_1.2,0.10198400020599366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,4,power_law_1.01,1.7486656188964844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,4,power_law_1.2,0.1542847990989685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,4,power_law_1.2,1.1682175636291503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,4,power_law_1.2,0.23176319599151612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,4,power_law_1.01,2.0584896087646483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,4,power_law_1.01,4.699654388427734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,4,power_law_1.2,0.29898240566253664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,4,power_law_1.2,1.2022527694702148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,4,power_law_1.01,5.3875774383544925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,4,power_law_1.2,0.3707072019577026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,4,power_law_1.01,2.4722496032714845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,4,power_law_1.01,0.3534656047821045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,4,power_law_1.2,1.2615424156188966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,4,power_law_1.2,0.44391679763793945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,4,power_law_1.01,0.3444799900054932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,4,power_law_1.2,1.1723199844360352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,4,power_law_1.01,3.112812805175781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,4,power_law_1.2,1.1721023559570312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,4,power_law_1.01,6.545587158203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,4,power_law_1.2,0.41225600242614746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,4,power_law_1.2,1.1991616249084474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,4,1,balanced,0.06078400214513143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,4,1,balanced,0.08920533458391826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,4,1,balanced,0.16757865746816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,4,1,balanced,0.2823093334833781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,4,power_law_1.2,0.5077760219573975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,4,power_law_1.01,4.111916732788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,4,1,balanced,0.5027413368225098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,4,power_law_1.2,1.1851776123046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,4,power_law_1.2,0.5405695915222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,4,1,balanced,0.972927967707316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,4,power_law_1.2,0.6750976085662842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,4,1,balanced,1.357093334197998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,4,power_law_1.2,1.291603183746338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,4,power_law_1.2,0.6739647865295411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,4,power_law_1.2,0.5934783935546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,4,1,balanced,1.3596266110738118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,4,power_law_1.2,1.406214427947998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,4,1,balanced,1.3787040710449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,4,power_law_1.2,1.0784895896911622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,4,power_law_1.2,0.638150405883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,4,power_law_1.01,5.332012939453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,4,1,balanced,1.3829867045084636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,4,power_law_1.01,0.3617471933364868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,4,power_law_1.2,0.6966976165771485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,4,power_law_1.2,1.6422271728515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,4,1,balanced,1.3638985951741536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,4,power_law_1.01,0.40928001403808595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,4,1,balanced,1.3643946647644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,4,power_law_1.2,0.7290431976318359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,4,power_law_1.2,1.6529407501220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,4,power_law_1.01,0.4147327899932861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,4,power_law_1.2,0.7215295791625976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,4,power_law_1.01,0.45537281036376953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,4,power_law_1.2,0.05548800230026245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,4,power_law_1.2,0.7367680072784424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,4,power_law_1.2,1.8777088165283202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,4,power_law_1.2,0.08283519744873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,4,power_law_1.01,0.6201727867126465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,4,1,power_law_1.2,7.172953796386719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,4,power_law_1.2,1.093273639678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,4,power_law_1.2,0.7627840042114258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,4,power_law_1.2,0.11416959762573242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,4,power_law_1.01,7.860787200927734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,4,power_law_1.01,0.6314559936523437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,4,1,balanced,0.041749333341916404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,4,power_law_1.2,2.0334272384643555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,4,1,balanced,0.06022400160630544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,4,power_law_1.2,0.14466559886932373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,4,1,balanced,0.086709330479304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,4,power_law_1.2,0.783903980255127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,4,1,balanced,0.14710932970046997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,4,power_law_1.2,1.0862784385681152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,4,1,balanced,0.27011199792226154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,4,power_law_1.01,0.6913599967956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,4,power_law_1.2,0.18558720350265503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,4,power_law_1.2,0.7917695999145508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,4,power_law_1.2,1.0194239616394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,4,power_law_1.01,0.8880448341369629
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,4,power_law_1.2,0.2448960065841675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,4,power_law_1.2,2.3232831954956055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,4,power_law_1.2,0.9796863555908203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,4,power_law_1.2,0.25455360412597655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,4,power_law_1.2,0.9974335670471192
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,4,power_law_1.01,1.0582655906677245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,4,power_law_1.2,1.0134016036987306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,4,power_law_1.2,0.31641600131988523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,4,power_law_1.2,2.690790367126465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,4,power_law_1.2,1.1649663925170899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,4,1,balanced,1.3682400385538738
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,4,power_law_1.01,1.452883243560791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,4,power_law_1.2,0.32493441104888915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,4,1,balanced,0.5003519852956136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,4,1,balanced,1.3693119684855144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,4,1,balanced,0.8631626764933268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,4,power_law_1.2,1.1457152366638184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,4,power_law_1.2,0.3106303930282593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,4,1,balanced,0.8634453614552816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,4,power_law_1.01,9.8399169921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,4,1,balanced,1.3799146016438801
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,4,power_law_1.2,0.32364161014556886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,4,power_law_1.2,3.47496337890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,4,power_law_1.2,1.148140811920166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,4,power_law_1.01,2.586086463928223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,4,1,balanced,1.3870933850606282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,4,power_law_1.2,0.32791039943695066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,4,1,balanced,1.403637409210205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,4,power_law_1.2,1.1833151817321776
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,4,power_law_1.2,0.32211201190948485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,4,power_law_1.2,1.157907199859619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,4,1,balanced,1.7434293429056804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,4,power_law_1.2,0.3195519924163818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,4,power_law_1.2,1.4228480339050293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,4,1,balanced,1.7748212814331055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,4,power_law_1.2,4.170918273925781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,4,1,balanced,0.8016266822814941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,4,1,balanced,0.06425066788991292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,4,1,balanced,0.07682666679223378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,4,1,balanced,0.09868799646695454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,4,power_law_1.2,1.569696044921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,4,1,balanced,0.15431466698646545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,4,1,balanced,0.27907200654347736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,4,1,balanced,0.539354681968689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,4,1,balanced,0.7878719965616862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,4,power_law_1.2,1.8342592239379882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,4,1,balanced,0.789685328801473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,4,1,balanced,1.3068532943725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,4,power_law_1.2,0.37380480766296387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,4,1,balanced,1.3124480247497559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,4,power_law_1.2,2.0809856414794923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,4,power_law_1.2,0.3720832109451294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,4,1,balanced,1.8172906239827473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,4,1,balanced,0.8047573566436768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,4,1,balanced,1.3150826295216878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,4,1,balanced,0.7744800249735514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,4,power_law_1.2,0.3749887943267822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,4,1,balanced,1.3226719697316487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,4,1,balanced,1.8475839296976726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,4,1,balanced,0.7762719790140787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,4,power_law_1.2,2.7131135940551756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,4,power_law_1.2,0.41295361518859863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,4,1,balanced,0.7821493148803711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,4,1,balanced,1.328879992167155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,4,1,balanced,1.9520266850789387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,4,1,balanced,0.788645346959432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,4,1,balanced,1.3399252891540527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,4,power_law_1.2,0.42357120513916013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,4,1,balanced,0.7970293362935384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,4,power_law_1.2,1.2779647827148437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,4,1,balanced,0.807365338007609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,4,1,balanced,1.3521812756856282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,4,power_law_1.2,0.49236478805541994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,4,1,balanced,2.0574933687845864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,4,1,balanced,0.8213919798533121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,4,power_law_1.2,3.300806427001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,4,1,balanced,1.3646559715270996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,4,power_law_1.2,1.5354111671447754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,4,power_law_1.2,0.6173247814178466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,4,1,balanced,2.2969706853230796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,4,1,balanced,1.3830453554789226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,4,1,balanced,0.031178665657838184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,4,power_law_1.2,0.6997759819030762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,4,1,balanced,0.03769599894682566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,4,1,balanced,1.4205279350280762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,4,1,balanced,0.05914133290449778
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,4,1,balanced,0.08683733145395915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,4,power_law_1.2,1.7965120315551757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,4,1,balanced,0.14283733566602072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,4,1,balanced,0.25305066506067914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,4,power_law_1.2,0.7961023807525635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,4,1,balanced,1.453546682993571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,4,1,balanced,3.4112265904744468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,4,power_law_1.2,5.448281478881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,4,1,balanced,0.36665066083272296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,4,1,balanced,0.36660265922546387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,4,power_law_1.2,4.3868671417236325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,4,1,balanced,1.5167519251505535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,4,1,balanced,0.3697119951248169
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,4,1,balanced,0.37307198842366535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,4,power_law_1.2,0.9809856414794922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,4,power_law_1.2,2.3081279754638673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,4,1,balanced,0.37882665793100995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,4,1,balanced,1.070133368174235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,4,1,balanced,1.5825279553731282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,4,1,balanced,0.38489067554473877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,4,1,balanced,3.9526987075805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,4,1,balanced,0.39060266812642414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,4,1,balanced,1.0918827056884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,4,1,balanced,0.40213334560394287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,4,1,balanced,1.7179039319356282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,4,power_law_1.2,1.1459839820861817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,4,1,balanced,0.41230932871500653
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,4,1,balanced,0.4248053232828776
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,4,1,balanced,0.46795201301574707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,4,1,balanced,0.5119306643803915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,4,1,balanced,2.7810773849487305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,4,power_law_1.2,1.5028736114501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,4,1,balanced,0.5093226830164591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,4,power_law_1.2,5.484076690673828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,4,power_law_1.2,6.721107482910156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,4,1,balanced,5.463653564453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,4,1,balanced,0.5738933483759562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,4,1,balanced,0.5986773173014323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,4,1,power_law_1.01,0.060134398937225345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,4,1,balanced,3.0478506088256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,4,1,balanced,0.6923466523488363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,4,1,power_law_1.01,0.044998401403427125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,4,1,power_law_1.01,0.08609279990196228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,4,1,balanced,0.846671978632609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,4,1,power_law_1.01,0.06279039978981019
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,4,power_law_1.2,3.0594112396240236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,4,1,balanced,1.133354663848877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,4,1,power_law_1.01,0.1503424048423767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,4,1,balanced,0.9095306396484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,4,1,power_law_1.01,0.09045119881629944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,4,1,balanced,1.181167999903361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,4,power_law_1.2,2.788326454162598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,4,1,balanced,4.303850809733073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,4,1,power_law_1.01,0.25743999481201174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,4,1,power_law_1.01,0.14736640453338623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,4,1,balanced,1.780469258626302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,4,1,balanced,1.2773226896921794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,4,1,power_law_1.01,0.2456831932067871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,4,1,power_law_1.01,0.43089919090270995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,4,1,balanced,1.3732372919718425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,4,1,balanced,1.3782347043355305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,4,1,power_law_1.01,0.3395711898803711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,4,power_law_1.2,7.7710014343261715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,4,1,power_law_1.01,0.5913599967956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,4,1,balanced,1.5736427307128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,4,1,balanced,1.9934666951497395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,4,power_law_1.2,3.8102336883544923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,4,1,power_law_1.01,0.46636161804199217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,4,1,balanced,5.802560170491536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,4,1,power_law_1.01,0.7457664012908936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,4,1,balanced,2.308128039042155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,4,1,power_law_1.01,0.6015679836273193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,4,1,power_law_1.01,0.9277759552001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,4,1,balanced,2.7141545613606772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,4,1,power_law_1.01,0.971884822845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,4,power_law_1.2,4.700422286987305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,4,1,balanced,3.704335848490397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,4,1,power_law_1.01,0.9852479934692383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,4,1,balanced,8.183685302734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,4,1,power_law_1.2,0.0553600013256073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,4,1,power_law_1.2,0.08118399977684021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,4,1,power_law_1.2,0.043968001008033754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,4,1,power_law_1.01,1.07957763671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,4,1,power_law_1.2,0.05919359922409058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,4,1,power_law_1.2,0.15315200090408326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,4,1,power_law_1.01,0.5741568088531495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,4,power_law_1.2,10.557215881347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,4,1,power_law_1.2,0.08568959832191467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,4,1,power_law_1.01,1.1035008430480957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,4,1,power_law_1.01,0.5839104175567627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,4,1,power_law_1.2,0.14620800018310548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,4,1,power_law_1.01,1.1074048042297364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,4,1,power_law_1.01,0.6219136238098144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,4,1,power_law_1.2,0.21917440891265869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,4,1,power_law_1.01,0.6378111839294434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,4,1,power_law_1.2,0.30882558822631834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,4,1,power_law_1.01,1.1305600166320802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,4,1,power_law_1.2,0.06152960062026978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,4,1,power_law_1.01,0.6480319976806641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,4,1,power_law_1.2,0.07455999851226806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,4,1,power_law_1.2,0.44225921630859377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,4,1,power_law_1.01,1.1596480369567872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,4,1,power_law_1.2,0.2542399883270264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,4,1,power_law_1.2,0.0964031994342804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,4,1,power_law_1.01,0.6645503997802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,4,1,power_law_1.2,0.5764863967895508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,4,1,power_law_1.2,0.1513856053352356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,4,1,power_law_1.2,0.38645761013031005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,4,1,power_law_1.01,1.1941184043884276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,4,1,power_law_1.01,0.6813504219055175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,4,1,power_law_1.01,0.061894398927688596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,4,1,power_law_1.2,0.5595136165618897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,4,1,power_law_1.2,0.2491391897201538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,4,1,power_law_1.2,0.5455167770385743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,4,1,power_law_1.01,0.7031936168670654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,4,1,power_law_1.01,0.029267200827598573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,4,1,power_law_1.01,1.2404543876647949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,4,1,power_law_1.2,0.33755519390106203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,4,1,power_law_1.2,0.5749695777893067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,4,1,power_law_1.01,0.7095935821533204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,4,1,power_law_1.2,0.6459199905395507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,4,1,power_law_1.01,0.03687680065631867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,4,1,power_law_1.2,0.3969343900680542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,4,1,power_law_1.2,0.6115392208099365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,4,1,power_law_1.01,1.5095744132995605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,4,1,power_law_1.01,0.056576001644134524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,4,1,power_law_1.2,0.8960063934326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,4,1,power_law_1.2,0.5329279899597168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,4,1,power_law_1.2,0.6314688205718995
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,4,1,power_law_1.01,0.08535040020942689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,4,1,power_law_1.01,1.5961024284362793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,4,1,power_law_1.2,0.9552767753601075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,4,1,power_law_1.01,0.1314560055732727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,4,1,power_law_1.2,0.6442495822906494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,4,1,power_law_1.2,0.9289536476135254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,4,1,power_law_1.01,0.18346240520477294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,4,1,power_law_1.2,0.9388607978820801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,4,1,power_law_1.2,0.6586239814758301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,4,1,power_law_1.01,1.6691776275634767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,4,1,power_law_1.2,0.9659775733947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,4,1,power_law_1.01,0.2259200096130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,4,1,power_law_1.2,0.6789247989654541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,4,1,power_law_1.01,0.9343999862670899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,4,1,power_law_1.2,1.074828815460205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,4,1,power_law_1.01,0.27409279346466064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,4,1,power_law_1.2,1.0054207801818849
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,4,1,power_law_1.2,0.02990719974040985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,4,1,power_law_1.2,0.7012032032012939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,4,1,power_law_1.01,0.9777983665466309
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,4,1,power_law_1.01,0.2817280054092407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,4,1,power_law_1.2,1.0781951904296876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,4,1,power_law_1.2,0.03720960021018982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,4,1,power_law_1.2,0.732966423034668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,4,1,power_law_1.2,1.0473088264465331
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,4,1,power_law_1.01,0.291430401802063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,4,1,power_law_1.01,1.068838405609131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,4,1,power_law_1.2,0.057740801572799684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,4,1,power_law_1.2,1.0961600303649903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,4,1,power_law_1.2,0.9684800148010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,4,1,power_law_1.2,1.0436991691589355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,4,1,power_law_1.2,0.0871616005897522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,4,1,power_law_1.01,1.168723201751709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,4,1,power_law_1.2,1.1423935890197754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,4,1,power_law_1.2,0.12259199619293212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,4,1,power_law_1.2,1.0292672157287597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,4,1,power_law_1.2,1.0982336044311523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,4,1,power_law_1.01,1.8166271209716798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,4,1,power_law_1.01,0.07377920150756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,4,1,power_law_1.01,1.3565247535705567
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,4,1,power_law_1.2,0.17218559980392456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,4,1,power_law_1.2,1.1515711784362792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,4,1,power_law_1.2,1.1276991844177247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,4,1,power_law_1.2,1.115340805053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,4,1,power_law_1.01,0.09713280200958252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,4,1,power_law_1.2,0.2116096019744873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,4,1,power_law_1.01,1.5570879936218263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,4,1,power_law_1.01,0.15166079998016357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,4,1,power_law_1.2,1.197004795074463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,4,1,power_law_1.2,1.1852736473083496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,4,1,power_law_1.01,2.075935935974121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,4,1,power_law_1.2,1.1657855987548829
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,4,1,power_law_1.2,0.2668735980987549
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,4,1,power_law_1.01,0.30382719039916994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,4,1,power_law_1.01,0.2596544027328491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,4,1,power_law_1.2,1.2141632080078124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,4,1,power_law_1.01,0.31329920291900637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,4,1,power_law_1.01,1.9359167098999024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,4,1,power_law_1.2,0.2725183963775635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,4,1,power_law_1.2,1.2328448295593262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,4,1,power_law_1.01,2.3392831802368166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,4,1,power_law_1.01,0.32705280780792234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,4,1,power_law_1.2,0.2825920104980469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,4,1,power_law_1.2,1.566707229614258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,4,1,power_law_1.01,0.3415744066238403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,4,1,power_law_1.2,1.3563263893127442
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,4,1,power_law_1.2,0.29753599166870115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,4,1,power_law_1.01,2.3124416351318358
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,4,1,power_law_1.01,0.3677824020385742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,4,1,power_law_1.2,0.3121151924133301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,4,1,power_law_1.2,1.598316764831543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,4,1,power_law_1.2,1.463430404663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,4,1,power_law_1.01,0.3867136001586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,4,1,power_law_1.2,0.3261568069458008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,4,1,power_law_1.01,0.43521919250488283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,4,1,power_law_1.2,1.4268159866333008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,4,1,power_law_1.01,3.0752704620361326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,4,1,power_law_1.01,0.3752703905105591
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,4,1,power_law_1.2,0.3472383975982666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,4,1,power_law_1.2,1.744748878479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,4,1,power_law_1.01,0.534662389755249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,4,1,power_law_1.2,0.3693696022033691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,4,1,power_law_1.2,1.6166143417358398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,4,1,power_law_1.01,0.48363518714904785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,4,1,power_law_1.2,1.8922239303588868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,4,1,power_law_1.2,0.391539192199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,4,1,power_law_1.01,0.6094975948333741
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,4,1,power_law_1.2,0.43253121376037595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,4,1,power_law_1.01,2.8861312866210938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,4,1,power_law_1.01,3.845075225830078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,4,1,power_law_1.2,1.992812728881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,4,1,power_law_1.01,0.5849343776702881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,balanced,0.0352906659245491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,balanced,0.04765866696834564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,4,1,power_law_1.2,2.178783988952637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,balanced,0.06145066519578298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,balanced,0.10032000144322713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,balanced,0.17178666591644287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,4,1,power_law_1.2,1.705695915222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,balanced,0.30326932668685913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,4,1,power_law_1.01,0.7519423961639404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,balanced,0.3227306604385376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,balanced,0.3195626735687256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,4,1,power_law_1.01,0.42332801818847654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,4,1,power_law_1.2,2.378937530517578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,4,1,power_law_1.01,3.405574417114258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,4,1,power_law_1.01,0.756166410446167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,4,1,power_law_1.2,2.4473215103149415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,4,1,power_law_1.2,1.9272319793701171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,4,1,power_law_1.01,0.5442431926727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,4,1,power_law_1.01,0.9876799583435059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,4,1,power_law_1.01,0.9559871673583984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,4,1,power_law_1.2,3.132499122619629
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,4,1,power_law_1.2,0.554585599899292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,4,1,power_law_1.2,2.3885631561279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,4,1,power_law_1.01,1.1399295806884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,4,1,power_law_1.2,0.5023935794830322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,4,1,power_law_1.01,0.9872384071350098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,4,1,power_law_1.01,4.428768157958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,balanced,0.32867199182510376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,balanced,0.32953067620595294
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,4,1,power_law_1.2,0.6292863845825195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,balanced,0.3482453425725301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,4,1,power_law_1.01,1.5383551597595215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,balanced,0.35124798615773517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,balanced,0.35494399070739746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,4,1,power_law_1.2,0.6051455974578858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,balanced,0.3627786636352539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,4,1,power_law_1.2,3.9280704498291015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,balanced,0.36820268630981445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,balanced,0.03329599897066752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,balanced,0.3746986786524455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,balanced,0.03585066646337509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,balanced,0.054010664423306785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,balanced,0.3882186810175578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,balanced,0.0766133318344752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,balanced,0.11346667011578877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,balanced,0.46118398507436115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,balanced,0.18170666694641113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,4,1,power_law_1.2,2.999872016906738
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,4,1,power_law_1.01,2.139571189880371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,balanced,0.4792746702829997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,balanced,0.5164106686909994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,4,1,power_law_1.01,5.778291320800781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,balanced,0.5529066721598307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,balanced,0.043882668018341064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,balanced,0.04901333153247833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,balanced,0.6270293394724528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,balanced,0.05946666498978933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,balanced,0.07629333436489105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,4,1,power_law_1.01,1.0410304069519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,balanced,0.12533866365750632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,balanced,0.7042773564656576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,balanced,0.19406400124231973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,balanced,0.19359999895095825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,balanced,1.0244959990183513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,balanced,0.19522666931152344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,4,1,power_law_1.2,3.45401611328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,balanced,0.3288373351097107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,4,1,power_law_1.01,1.0343999862670898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,balanced,0.33262399832407635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,balanced,1.1772692998250325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,balanced,0.3360000054041545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,balanced,0.34065600236256915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,balanced,0.34540267785390216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,balanced,1.687669277191162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,4,1,power_law_1.01,1.0868032455444336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,balanced,0.3572106758753459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,balanced,0.36452798048655194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,balanced,0.3734560012817383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,balanced,0.39168532689412433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,balanced,2.1712800661722818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,4,1,power_law_1.01,1.102297592163086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,balanced,0.42732266585032147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,4,1,power_law_1.2,2.8090751647949217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,power_law_1.01,0.05842559933662415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,balanced,0.45770665009816486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,4,1,power_law_1.2,4.661407852172852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,balanced,0.5186293522516886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,power_law_1.01,0.09566079974174499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,4,1,power_law_1.01,1.1354175567626954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,balanced,0.5832000176111857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,power_law_1.01,0.12138240337371826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,4,1,power_law_1.2,0.7712768077850342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,balanced,0.8816373348236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,power_law_1.01,0.1450111985206604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,4,1,power_law_1.01,1.1431872367858886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,balanced,0.20633065700531006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,balanced,1.0046292940775554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,power_law_1.01,0.18993279933929444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,4,1,power_law_1.2,0.7811200141906738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,4,1,power_law_1.2,3.6905982971191404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,4,1,power_law_1.01,1.17324800491333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,power_law_1.01,0.20667519569396972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,balanced,1.4385333061218262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,4,1,power_law_1.2,1.012294387817383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,power_law_1.01,0.28275198936462403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,4,1,power_law_1.2,5.777568054199219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,4,1,power_law_1.01,1.3037887573242188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,power_law_1.01,0.2852992057800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,power_law_1.01,0.2989376068115234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,power_law_1.01,0.05498239994049072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,balanced,0.02755733331044515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,balanced,0.029333333174387615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,4,1,power_law_1.01,1.4224639892578126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,power_law_1.01,0.06928640007972717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,balanced,0.03338133295377096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,balanced,0.0498933345079422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,4,1,power_law_1.2,4.520057678222656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,balanced,0.06895466645558675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,balanced,0.20665067434310913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,balanced,0.10502933462460835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,power_law_1.01,0.08893439769744874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,balanced,0.19982399543126425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,balanced,0.1074026624361674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,balanced,0.2011573314666748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,balanced,0.10922132929166158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,balanced,0.22473067045211792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,balanced,0.11126933495203654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,power_law_1.01,0.10661120414733886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,balanced,0.11321600278218587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,4,1,power_law_1.01,1.6608383178710937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,balanced,0.11628799637158711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,balanced,0.1127893328666687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,power_law_1.01,0.12634240388870238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,balanced,0.11530133088429768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,balanced,0.11866133411725362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,balanced,0.11550933122634888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,balanced,1.8719679514567058
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,balanced,0.11776000261306763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,power_law_1.01,0.14137599468231202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,balanced,0.12479466199874878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,4,1,power_law_1.01,1.8527423858642578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,balanced,0.1490666667620341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,power_law_1.01,0.17003519535064698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,power_law_1.01,0.3005631923675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,power_law_1.01,0.1816704034805298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,balanced,2.735210736592611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,power_law_1.01,0.2989248037338257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,4,1,power_law_1.2,6.15173110961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,power_law_1.01,0.17434240579605104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,4,1,power_law_1.01,2.295462417602539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,power_law_1.01,0.3286463975906372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,power_law_1.01,0.17166080474853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,balanced,0.22985066970189413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,balanced,0.23030932744344076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,power_law_1.01,0.32880001068115233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,balanced,0.24036266406377158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,balanced,3.601312001546224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,balanced,0.2492213249206543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,power_law_1.01,0.3250688076019287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,power_law_1.01,0.3412224054336548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,4,1,power_law_1.01,2.7375616073608398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,4,1,power_law_1.2,1.1758784294128417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,power_law_1.01,0.3542335987091064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,power_law_1.01,0.05728639960289002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,power_law_1.01,0.35690879821777344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,power_law_1.01,0.44585599899291994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,4,1,power_law_1.2,1.5584447860717774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,power_law_1.01,0.47049598693847655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,power_law_1.01,0.20908160209655763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,4,1,power_law_1.2,7.764019012451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,4,1,power_law_1.01,3.6118400573730467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,power_law_1.01,0.5343103885650635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,balanced,0.25225599606831867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,power_law_1.01,0.21502079963684081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,balanced,0.26733332872390747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,balanced,0.3219573299090068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,power_law_1.01,0.21624960899353027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,power_law_1.01,0.603116798400879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,balanced,0.34250664710998535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,balanced,0.13795199990272522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,power_law_1.01,0.2282815933227539
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,4,1,power_law_1.2,2.364633560180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,balanced,0.24665600061416626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,power_law_1.01,0.7219264030456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,balanced,0.1681706706682841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,power_law_1.01,0.23247361183166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,power_law_1.01,0.07071999907493591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,power_law_1.2,0.060134398937225345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,power_law_1.01,0.24404480457305908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,power_law_1.01,0.0806335985660553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,power_law_1.01,0.8488191604614258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,power_law_1.2,0.09596160054206848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,power_law_1.01,0.25617280006408694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,4,1,power_law_1.01,4.487321472167968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,power_law_1.2,0.10100480318069457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,power_law_1.01,0.31706240177154543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,power_law_1.01,1.1204416275024414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,power_law_1.2,0.13645440340042114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,power_law_1.01,0.33765759468078616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,power_law_1.2,0.16373759508132935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,power_law_1.01,0.39258880615234376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,power_law_1.01,1.3667455673217774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,power_law_1.2,0.20668160915374756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,power_law_1.01,0.4424767971038818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,power_law_1.2,0.2530431985855103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,balanced,0.1832746664683024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,power_law_1.01,0.5681791782379151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,power_law_1.2,0.27032320499420165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,power_law_1.01,0.028857600688934327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,power_law_1.01,0.6735040187835694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,power_law_1.2,0.2752320051193237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,power_law_1.01,0.04469119906425476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,4,1,power_law_1.01,6.169081497192383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,power_law_1.2,0.2896575927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,power_law_1.01,0.051622402667999265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,power_law_1.01,0.9086976051330566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,power_law_1.2,0.309113597869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,power_law_1.01,0.06643199920654297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,power_law_1.2,0.30113921165466306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,power_law_1.01,1.1225215911865234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,power_law_1.01,0.06909440159797668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,power_law_1.2,0.3193919897079468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,power_law_1.01,0.06849279999732971
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,balanced,0.2060746749242147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,power_law_1.2,0.3362175941467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,power_law_1.01,0.08773120045661927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,power_law_1.01,2.0223871231079102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,power_law_1.01,1.593388843536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,power_law_1.2,0.3490303993225098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,power_law_1.01,0.09455360174179077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,power_law_1.01,0.12023680210113526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,power_law_1.2,0.35583360195159913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,power_law_1.01,0.09702399969100953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,power_law_1.01,0.1310912013053894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,power_law_1.2,0.36375041007995607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,power_law_1.01,0.09884160161018371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,power_law_1.01,1.9968576431274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,power_law_1.01,0.14313600063323975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,power_law_1.01,2.4514368057250975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,power_law_1.2,0.4569087982177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,4,1,power_law_1.01,7.797644805908203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,power_law_1.01,0.10500479936599731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,power_law_1.2,0.04926080107688904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,balanced,0.38258667786916095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,power_law_1.01,0.1651263952255249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,power_law_1.2,0.4772928237915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,balanced,0.4222773313522339
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,power_law_1.01,0.09745280146598816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,power_law_1.2,0.06645119786262513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,power_law_1.01,0.18228479623794555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,balanced,0.5060586531956991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,power_law_1.2,0.05803520083427429
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,power_law_1.01,0.10204800367355346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,power_law_1.2,0.5494783878326416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,power_law_1.2,0.07692800164222717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,balanced,0.5895786682764689
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,balanced,0.29996800422668457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,power_law_1.2,0.07125759720802308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,power_law_1.01,0.10487040281295776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,power_law_1.01,0.2878079891204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,balanced,0.34934401512145996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,power_law_1.2,0.09170560240745544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,power_law_1.2,0.6294079780578613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,power_law_1.2,0.07512959837913513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,power_law_1.01,0.10823040008544922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,power_law_1.01,0.285696005821228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,power_law_1.2,0.11426559686660767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,power_law_1.2,0.10736000537872314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,power_law_1.01,0.11033600568771362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,power_law_1.2,0.762227201461792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,power_law_1.01,0.2948992013931274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,power_law_1.2,0.13649920225143433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,power_law_1.01,0.11610239744186401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,power_law_1.2,0.12579840421676636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,power_law_1.01,0.2993599891662598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,power_law_1.01,0.12353279590606689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,power_law_1.2,0.9111616134643554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,power_law_1.2,0.16999679803848267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,power_law_1.2,0.12565120458602905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,power_law_1.01,0.1376512050628662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,power_law_1.2,0.17702399492263793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,power_law_1.2,0.17443840503692626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,power_law_1.01,0.17786879539489747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,power_law_1.2,1.1771583557128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,power_law_1.2,0.17290879487991334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,power_law_1.01,0.17499519586563111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,balanced,0.8377386728922526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,power_law_1.2,0.17824000120162964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,power_law_1.01,0.21637120246887206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,balanced,0.4968693256378174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,power_law_1.2,1.49932804107666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,balanced,1.0015520254770915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,balanced,0.6440159877141317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,power_law_1.01,0.2569216012954712
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,power_law_1.2,0.02807680070400238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,balanced,0.029338667790095013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,balanced,0.0415040006240209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,balanced,1.42523193359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,balanced,0.058730666836102806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,power_law_1.2,0.04351359903812409
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,power_law_1.01,0.3290112018585205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,balanced,0.09769599636395772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,balanced,0.18116267522176108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,balanced,0.3284426728884379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,power_law_1.2,0.04927360117435455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,power_law_1.2,2.012972831726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,power_law_1.01,0.42249598503112795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,balanced,0.34016533692677814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,power_law_1.01,0.31534719467163086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,balanced,1.854629357655843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,balanced,0.3420906861623128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,balanced,0.02536533276240031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,power_law_1.2,0.06093440055847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,balanced,0.029690665503342945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,balanced,0.3281760017077128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,balanced,0.0422986646493276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,power_law_1.01,0.327455997467041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,balanced,0.06758399804433186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,power_law_1.01,0.5889088153839112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,power_law_1.2,0.15763200521469117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,balanced,0.11346667011578877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,power_law_1.2,0.06710399985313416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,power_law_1.01,0.3327167987823486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,power_law_1.2,0.29241600036621096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,power_law_1.2,0.078438401222229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,power_law_1.01,1.0049920082092285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,power_law_1.2,2.705958366394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,power_law_1.2,0.20519039630889893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,power_law_1.2,0.0964031994342804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,power_law_1.2,0.21217920780181884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,power_law_1.2,0.0980288028717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,power_law_1.2,0.2180351972579956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,power_law_1.2,0.22607359886169434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,power_law_1.2,0.23980159759521485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,balanced,0.3267306685447693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,power_law_1.2,0.2447295904159546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,balanced,0.3409279982248942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,balanced,0.19209067026774088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,balanced,0.3380853335062663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,power_law_1.2,0.25980799198150634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,balanced,0.33843199412027997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,balanced,0.34223465124766034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,power_law_1.01,0.34927361011505126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,power_law_1.2,0.31919360160827637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,power_law_1.2,0.27815680503845214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,balanced,0.3471200068791707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,balanced,0.3489866654078166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,power_law_1.01,0.38264319896697996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,power_law_1.2,0.3450432062149048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,balanced,0.35707199573516846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,balanced,0.45028265317281085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,power_law_1.01,0.4474495887756348
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,power_law_1.2,0.10200320482254029
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,balanced,0.027450665831565857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,power_law_1.01,0.49042558670043945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,power_law_1.2,0.09759359955787658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,balanced,0.029359998802344005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,balanced,0.03357866654793421
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,balanced,0.048858667413393654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,balanced,0.07042133311430614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,power_law_1.2,0.09880959987640381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,balanced,0.1088853379090627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,balanced,0.10969600081443787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,balanced,0.22265599171320596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,balanced,0.11264533797899882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,power_law_1.2,0.10038399696350098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,balanced,0.11379200220108032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,balanced,0.11518399914105733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,power_law_1.2,0.10220799446105958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,balanced,0.11728533109029134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,balanced,0.12198399504025777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,power_law_1.2,0.10986239910125732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,power_law_1.2,0.29251201152801515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,power_law_1.2,0.11091840267181396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,power_law_1.2,0.32753279209136965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,power_law_1.2,0.4011392116546631
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,power_law_1.2,0.11388800144195557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,power_law_1.2,0.3100287914276123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,power_law_1.2,0.4608128070831299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,power_law_1.2,0.33213438987731936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,power_law_1.2,0.5828735828399658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,power_law_1.01,0.6191296100616455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,power_law_1.2,0.352294397354126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,balanced,0.22258667151133218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,balanced,0.20972265799840292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,power_law_1.2,0.6823296070098877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,balanced,0.2113866607348124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,power_law_1.01,0.7309311866760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,balanced,0.21717333793640137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,balanced,0.03370666752258936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,balanced,0.21700799465179443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,balanced,0.12409067153930664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,balanced,0.2167253295580546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,balanced,0.12752532958984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,power_law_1.2,0.9332159996032715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,balanced,0.22111467520395914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,power_law_1.01,0.9765888214111328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,balanced,0.2263466715812683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,balanced,0.22842667500178018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,balanced,0.23443732659022012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,balanced,0.3080693284670512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,power_law_1.2,1.1439167976379394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,power_law_1.2,0.12324479818344117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,balanced,0.31891733407974243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,power_law_1.01,1.1647744178771973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,balanced,0.34426132837931317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,power_law_1.2,0.13201279640197755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,balanced,0.369269331296285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,balanced,0.417738676071167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,power_law_1.2,0.14396799802780152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,power_law_1.2,1.6893375396728516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,balanced,0.47009066740671795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,power_law_1.01,1.6915712356567383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,power_law_1.2,0.364902400970459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,power_law_1.2,0.39726080894470217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,balanced,0.13900799552599588
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,balanced,0.14384532968203226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,power_law_1.2,2.084275245666504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,power_law_1.2,0.46408958435058595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,balanced,0.4601920048395793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,power_law_1.01,2.1442880630493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,balanced,0.48206400871276855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,power_law_1.01,0.025062400102615356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,balanced,0.5056266784667969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,power_law_1.01,0.039417600631713866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,balanced,0.5529333353042603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,balanced,0.6024853388468424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,power_law_1.01,0.057760000228881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,balanced,0.9113759994506836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,power_law_1.01,0.09630079865455628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,balanced,0.7060266335805258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,power_law_1.2,0.20289919376373292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,power_law_1.01,3.0514047622680662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,balanced,1.0134773254394531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,power_law_1.01,0.14190080165863037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,balanced,0.8070560296376547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,power_law_1.2,0.20536320209503173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,power_law_1.01,0.18440959453582764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,balanced,1.4344852765401204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,balanced,1.1430933475494385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,power_law_1.2,0.23863680362701417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,balanced,0.15481600165367126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,power_law_1.01,0.24971520900726318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,balanced,0.15798933307329813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,balanced,0.1688800056775411
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,power_law_1.2,0.285318398475647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,balanced,1.4827680587768555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,power_law_1.2,0.5306687831878663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,power_law_1.01,0.2684927940368652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,balanced,0.03908266623814901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,balanced,0.056234667698542275
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,power_law_1.2,0.3586816072463989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,balanced,0.07254933317502339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,power_law_1.01,0.25362560749053953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,power_law_1.01,0.023494400084018707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,power_law_1.2,0.6383999824523926
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,power_law_1.2,0.45136637687683107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,power_law_1.01,0.029145601391792297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,power_law_1.01,0.28275840282440184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,power_law_1.2,0.7905087947845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,power_law_1.01,0.043161600828170776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,power_law_1.01,0.28476800918579104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,power_law_1.2,0.6239424228668213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,power_law_1.01,0.06558079719543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,power_law_1.01,0.28833279609680174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,power_law_1.2,1.0399040222167968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,power_law_1.01,0.09742720127105713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,power_law_1.01,0.2879168033599854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,power_law_1.2,1.1388287544250488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,power_law_1.01,0.30480639934539794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,balanced,0.1832586725552877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,power_law_1.2,1.3045951843261718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,balanced,0.20198933283487955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,power_law_1.01,0.31639039516448975
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,balanced,0.25948800643285114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,balanced,0.26258132855097455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,power_law_1.01,3.868582534790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,power_law_1.01,0.3245311975479126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,balanced,0.5187946557998657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,balanced,0.41441067059834796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,power_law_1.01,0.33755519390106203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,power_law_1.2,1.7540096282958983
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,balanced,0.5833333333333334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,power_law_1.01,0.41918082237243653
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,balanced,0.7552800178527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,power_law_1.01,0.44686079025268555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,power_law_1.2,0.02536959946155548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,power_law_1.2,2.0992319107055666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,power_law_1.01,0.4791232109069824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,power_law_1.01,0.11628160476684571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,power_law_1.2,0.04048640131950378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,balanced,1.8527199427286785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,power_law_1.01,0.5411263942718506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,power_law_1.2,0.058208000659942624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,power_law_1.01,0.17111680507659913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,power_law_1.2,0.02396160066127777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,power_law_1.2,0.09400960206985473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,power_law_1.01,0.6331776142120361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,power_law_1.01,0.17681920528411865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,power_law_1.2,0.029254400730133058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,power_law_1.2,0.1337280035018921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,power_law_1.01,0.17297279834747314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,power_law_1.2,0.04345600008964538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,power_law_1.01,0.7401152133941651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,power_law_1.2,0.1699903964996338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,power_law_1.01,0.178220796585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,power_law_1.2,0.06236799955368042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,balanced,0.13019200166066489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,power_law_1.01,0.9513279914855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,power_law_1.01,0.18469120264053346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,power_law_1.2,0.24680960178375244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,power_law_1.2,0.08885759711265565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,power_law_1.01,0.18836480379104614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,power_law_1.01,0.024915200471878052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,power_law_1.2,0.264300799369812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,power_law_1.2,0.11136000156402588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,power_law_1.01,1.1511808395385743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,power_law_1.01,0.19457279443740844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,power_law_1.01,0.026284798979759216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,power_law_1.2,0.2656064033508301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,power_law_1.2,0.16673280000686647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,power_law_1.01,0.19735679626464844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,power_law_1.2,0.27292160987854003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,power_law_1.01,0.03225600123405457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,power_law_1.2,0.17589759826660156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,power_law_1.01,0.20471680164337158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,power_law_1.01,1.5761024475097656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,power_law_1.2,0.2852992057800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,power_law_1.2,0.16947200298309326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,power_law_1.01,0.04754559993743897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,power_law_1.2,3.0804927825927733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,power_law_1.2,0.28504960536956786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,power_law_1.2,0.17651840448379516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,power_law_1.01,0.059308797121047974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,power_law_1.2,0.28983678817749026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,power_law_1.2,0.1859455943107605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,power_law_1.01,0.0747648000717163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,power_law_1.01,2.0105472564697267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,power_law_1.2,0.2988672018051147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,power_law_1.2,0.1876736044883728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,power_law_1.01,0.09203199744224548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,power_law_1.2,0.3244352102279663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,power_law_1.2,0.19103360176086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,power_law_1.2,0.32629120349884033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,power_law_1.01,0.09711359739303589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,power_law_1.01,0.03199360072612763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,power_law_1.2,0.34101119041442873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,power_law_1.2,4.446764755249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,power_law_1.01,0.09908480048179627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,power_law_1.01,0.09961599707603455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,power_law_1.01,0.21557118892669677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,power_law_1.01,0.22410879135131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,power_law_1.01,0.10654079914093018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,power_law_1.01,0.2843071937561035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,power_law_1.01,0.10855040550231934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,power_law_1.01,0.31363840103149415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,power_law_1.01,0.11312639713287354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,power_law_1.01,0.3506880044937134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,power_law_1.01,0.11857919692993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,power_law_1.2,0.19864959716796876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,power_law_1.01,0.1278656005859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,power_law_1.01,0.3964735984802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,power_law_1.2,0.20832641124725343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,balanced,0.20314133167266846
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,power_law_1.01,0.13628159761428832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,balanced,0.20325332880020142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,power_law_1.01,0.47621760368347166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,power_law_1.2,0.21972479820251464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,power_law_1.2,0.4416639804840088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,power_law_1.01,0.1575999975204468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,power_law_1.2,0.22868480682373046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,power_law_1.01,0.15191680192947388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,power_law_1.2,0.45885438919067384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,power_law_1.2,0.3027008056640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,power_law_1.01,0.17656960487365722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,power_law_1.2,0.5108863830566406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,power_law_1.2,0.023635199666023253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,power_law_1.2,0.3221951961517334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,power_law_1.01,0.1839359998703003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,power_law_1.2,0.02592639923095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,power_law_1.2,0.5538112163543701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,power_law_1.2,0.3689856052398682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,power_law_1.01,0.21828479766845704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,power_law_1.2,0.032076799869537355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,power_law_1.2,0.40610561370849607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,power_law_1.01,0.23384320735931396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,power_law_1.2,0.6516479969024658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,power_law_1.2,0.044486400485038755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,power_law_1.01,0.2824575901031494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,power_law_1.2,0.4919104099273682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,power_law_1.2,0.05424000024795532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,power_law_1.2,0.7579135894775391
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,power_law_1.01,0.37326080799102784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,power_law_1.2,0.5716095924377441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,power_law_1.2,0.07331200242042542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,power_law_1.2,0.9695232391357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,power_law_1.2,0.0881663978099823
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,power_law_1.01,0.4566527843475342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,power_law_1.2,0.7465407848358154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,power_law_1.2,0.09374079704284669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,power_law_1.01,0.6278783798217773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,power_law_1.01,0.034041601419448855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,power_law_1.2,1.17391996383667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,power_law_1.2,0.09613440036773682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,power_law_1.2,0.916806411743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,power_law_1.01,0.05355520248413086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,power_law_1.2,0.09908480048179627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,power_law_1.2,0.033024001121521
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,power_law_1.01,0.9135680198669434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,power_law_1.01,0.06977919936180114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,power_law_1.2,0.10290559530258178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,balanced,0.2054133415222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,power_law_1.2,1.5962688446044921
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,power_law_1.2,0.1094208002090454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,power_law_1.01,0.5573247909545899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,power_law_1.2,0.1131775975227356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,power_law_1.2,0.1195520043373108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,power_law_1.01,0.7327104091644288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,power_law_1.2,2.016524887084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,power_law_1.2,0.13116159439086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,power_law_1.2,0.13889280557632447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,power_law_1.01,0.9013248443603515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,power_law_1.2,0.1605247974395752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,power_law_1.2,0.1573248028755188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,power_law_1.2,1.2585023880004882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,power_law_1.2,1.6155136108398438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,power_law_1.01,1.2399616241455078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,power_law_1.01,1.583097553253174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,power_law_1.2,0.03537920117378235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,power_law_1.01,0.09624959826469422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,power_law_1.2,0.05363199710845947
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,power_law_1.2,0.18110079765319825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,power_law_1.2,0.0679744005203247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,power_law_1.2,0.18806400299072265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,power_law_1.2,0.2217024087905884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,power_law_1.2,0.23841280937194825
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,power_law_1.2,0.2880192041397095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,power_law_1.2,0.3745728015899658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,power_law_1.2,0.4656703948974609
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,power_law_1.2,0.6360576152801514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,balanced,0.34562134742736816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,balanced,0.3476746479670207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,power_law_1.2,0.9706175804138184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,balanced,0.34942400455474854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,balanced,0.3495519955952962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,balanced,0.351642648379008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,balanced,0.3583466609319051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,balanced,0.36208534240722656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,balanced,0.36562665303548175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,balanced,0.3739253282546997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,balanced,0.39189334710439044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,balanced,0.40998931725819904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,balanced,0.4407680034637451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,balanced,0.4740639925003052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,balanced,0.7526240348815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,balanced,0.8163466453552246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,power_law_1.01,0.12828799486160278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,balanced,1.1652320226033528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,power_law_1.01,0.15550719499588012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,power_law_1.2,0.08984320163726807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,4,power_law_1.01,0.05318400263786316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,power_law_1.01,0.16399359703063965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,4,power_law_1.01,0.08845440149307252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,4,power_law_1.01,0.08954880237579346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,4,power_law_1.01,0.09994879961013795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,4,power_law_1.01,0.13638399839401244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,4,balanced,0.037605332831541695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,4,balanced,0.041509332756201424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,4,power_law_1.01,0.15973119735717772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,4,power_law_1.01,0.16102399826049804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,4,power_law_1.01,0.17260160446166992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,4,power_law_1.01,0.18152960538864135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,balanced,1.5189332962036133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,4,power_law_1.01,0.1910912036895752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,4,power_law_1.01,0.19340800046920775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,balanced,2.2097973823547363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,4,power_law_1.01,0.1886016011238098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,4,balanced,0.033728001018365227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,4,balanced,0.03746666759252548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,4,balanced,0.04264533519744873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,4,power_law_1.01,0.19001599550247192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,4,power_law_1.01,0.03880960047245026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,4,power_law_1.01,0.19822720289230347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,4,balanced,0.05597866574923197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,4,balanced,0.09360532959302266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,4,power_law_1.01,0.056428802013397214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,balanced,2.930037180582682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,4,balanced,0.14549332857131958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,4,power_law_1.01,0.20971519947052003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,4,balanced,0.1781866749127706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,4,power_law_1.01,0.057043200731277464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,4,power_law_1.01,0.21351680755615235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,4,power_law_1.01,0.07147520184516906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,4,power_law_1.01,0.22885758876800538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,4,power_law_1.01,0.08178560137748718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,4,power_law_1.01,0.2756351947784424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,4,power_law_1.01,0.1025920033454895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,4,power_law_1.01,0.3073728084564209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,4,power_law_1.01,0.11751680374145508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,4,power_law_1.01,0.36080639362335204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,4,power_law_1.01,0.1164736032485962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,4,power_law_1.01,0.12899199724197388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,4,power_law_1.01,0.4056447982788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,4,power_law_1.01,0.12226560115814208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,4,power_law_1.01,0.5225152015686035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,4,power_law_1.01,0.13126399517059326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,4,balanced,0.19978666305541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,4,balanced,0.1987946629524231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,4,power_law_1.01,0.13007359504699706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,4,balanced,0.208570659160614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,power_law_1.2,0.12428159713745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,4,power_law_1.01,0.6433599948883056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,4,balanced,0.20790932575861612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,4,power_law_1.01,0.13617279529571533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,4,balanced,0.20712000131607056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,4,balanced,0.20592532555262247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,power_law_1.2,0.15371520519256593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,4,balanced,0.20722132921218872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,4,power_law_1.01,0.8547712326049804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,4,balanced,0.21134400367736816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,4,balanced,0.2137333353360494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,power_law_1.2,0.1623744010925293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,4,balanced,0.21951999266942343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,4,balanced,0.2304426630338033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,4,balanced,0.06730133295059204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,4,balanced,0.27405333518981934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,4,power_law_1.01,1.1042304039001465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,4,balanced,0.29050666093826294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,4,balanced,0.3242986599604289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,4,balanced,0.3553440173467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,4,balanced,0.5104053417841593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,4,power_law_1.01,1.5030719757080078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,4,balanced,0.5734560092290243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,4,power_law_1.2,0.05305600166320801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,4,power_law_1.01,0.14451199769973755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,4,power_law_1.2,0.0705024003982544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,4,power_law_1.01,2.046022415161133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,4,power_law_1.01,0.14967679977416992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,4,power_law_1.2,0.08833919763565064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,4,power_law_1.01,0.15808639526367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,4,power_law_1.2,0.10744320154190064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,4,balanced,0.09993066390355428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,4,power_law_1.01,0.17184000015258788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,4,balanced,0.11709866921106975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,4,power_law_1.2,0.12171519994735717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,4,power_law_1.01,0.2059648036956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,4,power_law_1.2,0.15518720149993898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,4,power_law_1.01,0.22839040756225587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,4,power_law_1.01,0.28316800594329833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,power_law_1.01,0.28591361045837405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,4,power_law_1.01,0.32673919200897217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,power_law_1.01,0.28792319297790525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,4,power_law_1.01,0.43131518363952637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,power_law_1.01,0.29989759922027587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,4,power_law_1.01,0.5189568042755127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,power_law_1.01,0.2999232053756714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,4,power_law_1.01,0.7158143997192383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,power_law_1.01,0.30572800636291503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,power_law_1.01,0.31296639442443847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,4,power_law_1.01,0.9039999961853027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,power_law_1.01,0.3200576066970825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,4,power_law_1.2,0.03993600010871887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,power_law_1.01,0.3282815933227539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,4,power_law_1.01,1.2986623764038085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,4,power_law_1.2,0.05246719717979431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,power_law_1.01,0.35105280876159667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,4,power_law_1.2,0.060864001512527466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,power_law_1.01,0.39344000816345215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,4,power_law_1.2,0.07041280269622803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,4,1,balanced,0.025605333348115284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,4,1,balanced,0.03589866558710734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,4,1,balanced,0.054058666030565895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,4,balanced,0.13462400436401367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,power_law_1.01,0.4405055999755859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,4,balanced,0.7984906832377116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,4,balanced,0.13306666413942972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,4,balanced,0.9326879978179932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,power_law_1.01,0.5238207817077637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,4,balanced,1.370805263519287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,power_law_1.01,0.6029759883880615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,4,balanced,1.7308799425760906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,power_law_1.01,0.7841856002807617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,4,1,balanced,0.025061334172884624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,4,1,balanced,0.02920000006755193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,4,1,balanced,0.03719466676314672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,4,1,balanced,0.06419200201829274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,4,1,balanced,0.09893866380055745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,4,1,balanced,0.11986133456230164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,power_law_1.01,0.9440447807312011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,4,1,balanced,0.1402079959710439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,4,1,balanced,0.13961066802342734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,4,power_law_1.01,1.6871679306030274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,4,1,balanced,0.13320533434549967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,4,1,balanced,0.0929813285668691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,4,1,balanced,0.13380799690882364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,4,1,balanced,0.15474133690198263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,4,1,balanced,0.13872533043225607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,4,1,balanced,0.19539199272791544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,4,1,balanced,0.13665067156155905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,power_law_1.01,1.2908479690551757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,4,1,balanced,0.1386666695276896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,4,power_law_1.2,0.15935360193252562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,4,1,balanced,0.14148799578348795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,4,1,balanced,0.1434399982293447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,4,power_law_1.2,0.1734336018562317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,power_law_1.01,1.6127168655395507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,power_law_1.2,0.27178239822387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,power_law_1.2,0.287718391418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,power_law_1.01,2.266035270690918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,4,power_law_1.2,0.08068479895591736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,power_law_1.2,0.2967103958129883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,4,power_law_1.2,0.0995519995689392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,4,power_law_1.2,0.11537280082702636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,power_law_1.2,0.3031872034072876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,4,1,balanced,0.1470026671886444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,4,power_law_1.2,0.12095999717712402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,4,1,balanced,0.15236266454060873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,power_law_1.2,0.30353920459747313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,4,1,balanced,0.1914880077044169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,4,power_law_1.2,0.12257920503616333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,4,1,balanced,0.1997119983037313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,4,1,balanced,0.22155733903249106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,power_law_1.2,0.3174976110458374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,power_law_1.01,2.909619140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,4,power_law_1.2,0.1211967945098877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,4,1,balanced,0.24293333292007446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,4,1,balanced,0.35969066619873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,power_law_1.2,0.3226367950439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,4,power_law_1.2,0.13313920497894288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,4,1,power_law_1.01,0.02255360037088394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,4,1,balanced,0.39638932545979816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,4,1,power_law_1.01,0.02876800000667572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,4,power_law_1.2,0.1388543963432312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,4,1,balanced,0.564240018526713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,4,balanced,0.13303466637929282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,4,balanced,0.1346773306528727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,4,1,power_law_1.01,0.03937279880046844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,4,power_law_1.2,0.13653119802474975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,4,1,balanced,0.6504533290863037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,4,1,power_law_1.01,0.057760000228881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,4,power_law_1.2,0.1443071961402893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,4,1,balanced,0.9752586682637533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,4,1,balanced,0.2092693249384562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,4,1,power_law_1.01,0.0707264006137848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,4,1,balanced,0.20982933044433594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,4,power_law_1.2,0.15111039876937865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,4,1,power_law_1.01,0.1014143943786621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,4,1,balanced,1.2244853178660076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,4,power_law_1.2,0.1512768030166626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,4,1,power_law_1.01,0.11811840534210205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,4,power_law_1.2,0.19441280364990235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,4,1,power_law_1.2,0.023206399381160737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,4,power_law_1.2,0.17191040515899658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,4,1,power_law_1.01,0.11998080015182495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,4,1,power_law_1.2,0.03594880104064942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,4,power_law_1.2,0.21136000156402587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,4,1,power_law_1.01,0.12080639600753784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,4,1,power_law_1.2,0.052179199457168576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,4,power_law_1.2,0.23427200317382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,4,1,power_law_1.01,0.1245311975479126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,4,1,power_law_1.2,0.08073599934577942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,power_law_1.2,0.33836801052093507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,4,power_law_1.2,0.2899967908859253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,4,1,power_law_1.01,0.12915199995040894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,4,1,power_law_1.2,0.09288319945335388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,4,1,power_law_1.01,0.1279039978981018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,4,1,power_law_1.2,0.15561599731445314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,4,power_law_1.2,0.34087040424346926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,power_law_1.2,0.37125120162963865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,4,1,power_law_1.01,0.02332800030708313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,4,1,power_law_1.01,0.13107199668884278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,4,1,power_law_1.2,0.17112959623336793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,4,power_law_1.2,0.44323201179504396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,power_law_1.2,0.41880321502685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,4,1,power_law_1.01,0.03436160087585449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,4,1,power_law_1.2,0.17598719596862794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,4,1,power_law_1.01,0.05203199982643127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,4,1,power_law_1.2,0.17858560085296632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,4,power_law_1.2,0.5516352176666259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,power_law_1.2,0.45302400588989256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,4,1,power_law_1.01,0.0835968017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,4,1,power_law_1.2,0.18065919876098632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,4,power_law_1.2,0.19140479564666749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,4,1,power_law_1.2,0.19617279767990112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,4,1,power_law_1.01,0.10640640258789062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,power_law_1.2,0.5342144012451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,4,1,power_law_1.2,0.19477759599685668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,4,1,power_law_1.01,0.16000640392303467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,power_law_1.2,0.6231552124023437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,4,1,power_law_1.2,0.19880319833755494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,4,1,power_law_1.2,0.20131199359893798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,power_law_1.2,0.7956480026245117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,4,1,power_law_1.2,0.2030719995498657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,4,1,power_law_1.2,0.20880639553070068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,4,1,power_law_1.01,0.13689600229263305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,4,1,balanced,0.19435733556747437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,power_law_1.2,0.9587648391723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,4,1,power_law_1.2,0.2152640104293823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,4,1,balanced,0.19402666886647543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,4,1,power_law_1.01,0.13850879669189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,4,1,power_law_1.2,0.2693120002746582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,4,1,power_law_1.01,0.1435520052909851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,4,1,power_law_1.2,0.3001919984817505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,4,1,power_law_1.01,0.15032320022583007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,power_law_1.2,1.283033561706543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,4,1,power_law_1.01,0.18887679576873778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,4,1,power_law_1.2,0.3321536064147949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,4,1,power_law_1.01,0.20628480911254882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,4,1,power_law_1.2,0.3787519931793213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,4,1,power_law_1.01,0.24407680034637452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,power_law_1.2,1.6134208679199218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,4,1,power_law_1.2,0.4583295822143555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,4,1,power_law_1.01,0.2786303997039795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,4,1,power_law_1.2,0.5420351982116699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,power_law_1.2,2.2604352951049806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,4,balanced,0.13834666212399802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,4,1,power_law_1.2,0.719974422454834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,4,1,power_law_1.2,0.8934720039367676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,4,1,power_law_1.01,0.1741312026977539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,4,1,power_law_1.01,0.17433600425720214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,4,1,power_law_1.01,0.3495232105255127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,power_law_1.2,2.9136959075927735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,4,1,power_law_1.01,0.4213696002960205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,4,1,power_law_1.2,0.023315200209617616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,4,power_law_1.2,0.19121919870376586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,4,1,power_law_1.01,0.5635968208312988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,4,1,power_law_1.2,0.030502399802207945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,4,power_law_1.2,0.18922239542007446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,4,1,power_law_1.2,0.03826560080051422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,4,1,power_law_1.01,0.7055615901947021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,4,power_law_1.2,0.1971392035484314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,4,1,power_law_1.2,0.05454080104827881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,4,power_law_1.2,0.20424959659576417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,4,1,power_law_1.2,0.06883199810981751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,4,1,power_law_1.01,0.9991999626159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,4,power_law_1.2,0.21281280517578124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,4,1,power_law_1.2,0.10247679948806762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,4,1,balanced,0.20812267065048218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,4,1,power_law_1.2,1.249516773223877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,4,power_law_1.2,0.21626880168914794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,4,1,balanced,0.2046026587486267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,4,1,balanced,0.20500266551971436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,4,power_law_1.2,0.23721599578857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,4,1,balanced,0.20974934101104736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,4,1,power_law_1.01,1.287283229827881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,4,1,balanced,0.20940266052881876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,4,1,balanced,0.21144533157348633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,4,balanced,0.10082133611043294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,4,1,balanced,0.21894933780034384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,4,power_law_1.2,0.7493760108947753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,4,balanced,0.15849066774050394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,4,balanced,0.26768000920613605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,4,balanced,0.49297598997751874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,4,balanced,0.9077706336975098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,4,balanced,1.2646079858144124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,4,balanced,1.3028053442637126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,4,1,power_law_1.2,0.11831680536270142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,4,1,power_law_1.01,0.179257595539093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,4,balanced,0.14195199807484946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,4,balanced,0.14435199896494547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,4,1,power_law_1.2,0.1236672043800354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,4,1,power_law_1.01,0.18343679904937743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,4,balanced,1.3093706766764324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,4,1,power_law_1.2,0.12376960515975952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,4,balanced,1.4047412872314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,4,1,power_law_1.2,0.12401280403137208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,4,1,power_law_1.2,0.13237760066986085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,4,balanced,1.4118986129760742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,4,1,power_law_1.2,0.13076479434967042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,4,1,power_law_1.2,0.13254400491714477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,4,balanced,1.4738346735636394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,4,1,power_law_1.2,0.13703680038452148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,4,1,power_law_1.2,0.1400063991546631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,4,power_law_1.2,0.9200832366943359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,4,1,power_law_1.2,0.1440127968788147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,4,balanced,0.15250133474667868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,4,balanced,0.15837333599726358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,4,1,power_law_1.2,0.15041919946670532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,4,balanced,0.16270933548609415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,4,balanced,0.1752906640370687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,4,1,power_law_1.2,0.1931071996688843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,4,1,power_law_1.2,0.21468160152435303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,4,1,power_law_1.2,0.2483520030975342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,4,1,power_law_1.2,0.28548479080200195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,4,1,balanced,0.2670026620229085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,4,1,power_law_1.2,0.3523904085159302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,4,1,balanced,0.27534933884938556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,4,1,balanced,0.2934559981028239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,4,balanced,1.4925333658854167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,4,1,balanced,0.3142079909642537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,4,1,balanced,0.46909864743550617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,4,balanced,1.5119786262512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,4,1,balanced,0.5036426782608032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,4,1,balanced,0.7142879962921143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,4,balanced,1.5376639366149902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,4,1,balanced,0.7980319658915201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,4,balanced,1.5677439371744792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,4,1,balanced,1.2009387016296387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,4,power_law_1.2,0.275929594039917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,4,balanced,1.6116479237874348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,4,1,balanced,1.4861119588216145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,4,power_law_1.2,0.3035007953643799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,4,1,power_law_1.2,0.4212672233581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,4,balanced,1.6694025993347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,4,power_law_1.2,0.37041919231414794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,4,1,power_law_1.2,0.5643968105316162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,4,1,power_law_1.01,0.19111679792404174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,4,power_law_1.2,0.4278207778930664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,4,balanced,1.9008852640787761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,4,1,power_law_1.01,0.1900480031967163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,4,power_law_1.2,0.5630335807800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,4,1,power_law_1.01,0.18663040399551392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,4,1,power_law_1.01,0.19376640319824218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,4,balanced,1.980218728383382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,4,1,power_law_1.01,0.20383360385894775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,4,1,power_law_1.01,0.2054975986480713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,4,balanced,2.1808053652445474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,4,1,power_law_1.01,0.21283841133117676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,4,balanced,0.2061226765314738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,4,balanced,0.22593067089716592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,4,balanced,0.2603466709454854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,4,balanced,2.314623991648356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,4,balanced,0.2958986759185791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,4,balanced,0.4189866781234741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,4,balanced,0.4877599875132243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,4,1,power_law_1.2,0.7065983772277832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,4,power_law_1.2,0.6572480201721191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,4,balanced,3.558490753173828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,4,balanced,0.07086400190989177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,4,balanced,0.10322133700052898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,4,balanced,0.16788800557454428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,4,balanced,3.849546750386556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,4,power_law_1.2,1.3574336051940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,4,balanced,5.484240214029948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,4,1,power_law_1.01,0.26527359485626223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,4,balanced,6.557472229003906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,4,1,power_law_1.01,0.28159360885620116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,4,balanced,0.29053332408269245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,4,balanced,0.5028746525446574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,4,1,power_law_1.01,0.3263360023498535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,4,balanced,0.701141357421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,4,1,power_law_1.01,0.3740031957626343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,4,1,power_law_1.01,0.4533504009246826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,4,power_law_1.2,1.6959871292114257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,4,power_law_1.2,0.9030143737792968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,4,1,power_law_1.01,0.5353407859802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,4,balanced,0.6806720097859701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,4,balanced,0.8240853150685629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,4,1,power_law_1.01,0.7099520206451416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,4,balanced,1.2066506544748943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,4,1,power_law_1.01,0.88023681640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,4,power_law_1.2,1.1243391990661622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,4,balanced,9.964842478434244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,4,balanced,1.5436426798502605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,4,1,power_law_1.01,1.2303680419921874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,4,power_law_1.2,1.6947904586791993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,4,balanced,0.08052800099054973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,4,power_law_1.01,0.25626239776611326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,4,balanced,0.1090773344039917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,4,balanced,0.17722133795420328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,4,balanced,0.3094453414281209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,4,1,power_law_1.01,1.5884672164916993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,4,balanced,0.5245813528696696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,4,power_law_1.01,0.4555647850036621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,4,balanced,0.7258933385213217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,4,balanced,0.7354346911112467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,4,power_law_1.01,0.15397119522094727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,4,power_law_1.01,0.45964798927307127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,4,balanced,0.7440746625264486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,4,balanced,0.8030880292256674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,4,balanced,1.2978506882985432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,4,1,power_law_1.2,0.9947327613830567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,4,power_law_1.01,0.2604543924331665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,4,power_law_1.01,0.6355648040771484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,4,balanced,0.8075679937998453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,4,balanced,1.3083840211232503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,4,1,power_law_1.2,1.6058048248291015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,4,power_law_1.01,0.27239038944244387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,4,balanced,1.3305493195851643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,4,power_law_1.01,0.7454207897186279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,4,balanced,12.6156857808431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,4,balanced,1.350229263305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,4,balanced,1.3761547406514485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,4,power_law_1.01,1.2162752151489258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,4,power_law_1.2,0.2596992015838623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,4,balanced,1.4120319684346516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,4,power_law_1.2,0.3415168046951294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,4,power_law_1.01,1.1209792137145995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,4,balanced,1.4471786816914876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,4,power_law_1.2,0.49509119987487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,4,balanced,1.4866560300191243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,4,power_law_1.01,1.1802623748779297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,4,power_law_1.01,0.3636224031448364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,4,balanced,1.5582985877990723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,4,power_law_1.2,0.6465407848358155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,4,power_law_1.01,1.201375961303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,4,balanced,1.6922346750895183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,4,power_law_1.2,0.7531904220581055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,4,power_law_1.01,1.201638412475586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,4,balanced,1.8257333437601726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,4,power_law_1.2,1.0527999877929688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,4,power_law_1.01,1.371673583984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,4,balanced,2.920543988545736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,4,power_law_1.2,1.1399999618530274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,4,power_law_1.01,1.3491071701049804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,4,1,power_law_1.2,1.283187198638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,4,power_law_1.01,0.45224318504333494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,4,balanced,3.1819680531819663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,4,balanced,0.7550133069356283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,4,power_law_1.2,1.0991104125976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,4,power_law_1.01,1.3983488082885742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,4,balanced,0.7610560258229574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,4,power_law_1.01,0.6301119804382325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,4,power_law_1.2,1.235212802886963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,4,power_law_1.01,0.7323008060455323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,4,power_law_1.01,1.4279423713684083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,4,balanced,4.578794797261556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,4,power_law_1.01,0.6855743885040283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,4,power_law_1.01,1.4888383865356445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,4,power_law_1.01,0.669382381439209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,4,power_law_1.01,0.16759040355682372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,4,power_law_1.01,0.6726975917816163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,4,balanced,5.0943145751953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,4,power_law_1.01,0.2855168104171753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,4,power_law_1.01,0.7768896102905274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,4,power_law_1.01,0.2728832006454468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,4,power_law_1.01,0.8080575942993165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,4,power_law_1.2,1.2400256156921388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,4,power_law_1.01,0.37751040458679197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,4,power_law_1.01,0.8337023735046387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,4,power_law_1.01,0.4745791912078857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,4,power_law_1.2,1.3272319793701173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,4,power_law_1.01,0.8720000267028809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,4,power_law_1.01,1.5047807693481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,4,balanced,7.848997116088867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,4,power_law_1.01,0.6359807968139648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,4,power_law_1.2,0.15280640125274658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,4,power_law_1.2,1.3677375793457032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,4,power_law_1.01,0.8974271774291992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,4,power_law_1.2,0.19994239807128905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,4,power_law_1.01,1.599123191833496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,4,power_law_1.01,0.6462592124938965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,4,power_law_1.01,0.9408255577087402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,4,power_law_1.2,1.42926082611084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,4,power_law_1.2,0.2835520029067993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,4,power_law_1.01,0.6798655986785889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,4,power_law_1.01,0.9972736358642578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,4,power_law_1.01,1.895903968811035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,4,power_law_1.2,1.4526975631713868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,4,power_law_1.2,0.36469120979309083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,4,power_law_1.01,1.1869952201843261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,4,power_law_1.01,1.2024959564208983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,4,power_law_1.2,0.4551743984222412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,4,power_law_1.01,2.056742477416992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,4,power_law_1.2,1.5322175979614259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,4,power_law_1.01,1.1457535743713378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,4,power_law_1.01,1.3358912467956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,4,balanced,9.732597351074219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,4,power_law_1.2,0.6477183818817138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,4,power_law_1.01,1.1989631652832031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,4,power_law_1.2,1.5919936180114747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,4,power_law_1.01,1.5714303970336914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,4,power_law_1.01,2.3400447845458983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,4,power_law_1.2,0.7352831840515137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,4,power_law_1.01,1.2178175926208497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,4,power_law_1.2,0.6823808193206787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,4,power_law_1.2,1.6882047653198242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,4,power_law_1.01,1.823379135131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,4,power_law_1.01,1.2676223754882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,4,power_law_1.2,0.6586624145507812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,4,power_law_1.01,2.7438079833984377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,4,balanced,0.8549493153889974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,4,power_law_1.2,0.691321611404419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,4,power_law_1.2,1.9707199096679688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,4,power_law_1.01,1.361734390258789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,4,power_law_1.01,2.347929573059082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,4,power_law_1.2,0.8126976013183593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,4,power_law_1.01,1.3900032043457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,4,power_law_1.01,3.451071929931641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,4,power_law_1.2,0.8273856163024902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,4,power_law_1.2,2.130771255493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,4,power_law_1.01,2.859654426574707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,4,power_law_1.2,0.8607232093811035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,4,balanced,14.396320343017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,4,power_law_1.01,1.4866623878479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,4,power_law_1.2,0.8813119888305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,4,power_law_1.2,2.4592512130737303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,4,power_law_1.01,1.6111616134643554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,4,power_law_1.01,4.094937515258789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,4,power_law_1.2,0.9204031944274902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,4,power_law_1.01,3.840108871459961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,4,power_law_1.01,1.8789247512817382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,4,power_law_1.2,2.8634944915771485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,4,power_law_1.2,0.9715392112731933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,4,power_law_1.2,0.9996288299560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,4,power_law_1.01,2.1915456771850588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,4,power_law_1.01,5.551615905761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,4,power_law_1.2,1.2373056411743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,4,power_law_1.2,3.6388481140136717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,4,power_law_1.01,4.997862243652344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,4,power_law_1.01,2.873119926452637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,4,power_law_1.2,1.3859904289245606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,4,balanced,0.8771733442942301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,4,balanced,0.8953279654184977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,4,power_law_1.2,1.607731246948242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,4,balanced,0.9230720202128092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,4,power_law_1.2,4.245228958129883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,4,balanced,18.17131169637044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,4,balanced,0.9457813103993734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,4,power_law_1.01,3.3784320831298826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,4,balanced,0.9737333456675211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,4,power_law_1.2,1.8168767929077148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,4,balanced,1.0272479852040608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,4,power_law_1.01,7.296383666992187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,4,power_law_1.01,6.839603424072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,4,power_law_1.2,2.3801536560058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,4,power_law_1.01,4.596563339233398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,4,power_law_1.2,5.912614440917968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,4,power_law_1.2,2.2199935913085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,4,power_law_1.2,0.16792960166931153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,4,power_law_1.2,2.9437887191772463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,4,power_law_1.2,0.2357952117919922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,4,power_law_1.2,0.2715519905090332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,4,power_law_1.01,5.704838562011719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,4,power_law_1.2,0.3573632001876831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,4,power_law_1.2,4.0023681640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,4,power_law_1.01,9.002054595947266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,4,power_law_1.01,10.189055633544921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,4,power_law_1.2,7.424275207519531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,4,power_law_1.2,5.059276962280274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,4,balanced,1.1987733046213787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,4,power_law_1.2,0.4701375961303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,4,power_law_1.01,8.270553588867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,4,balanced,1.2865599791208904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,4,balanced,1.4554880460103352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,4,power_law_1.2,0.6605567932128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,4,balanced,1.6229759852091472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,4,power_law_1.2,0.6611775875091552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,4,power_law_1.2,10.740480041503906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,4,balanced,2.344618638356527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,4,1,balanced,0.06681600213050842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,4,1,balanced,0.09516800443331401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,4,1,balanced,0.1495786706606547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,4,1,balanced,0.2748639980951945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,4,power_law_1.01,13.105235290527343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,4,1,balanced,0.4994399944941203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,4,balanced,2.683050791422526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,4,1,balanced,0.7159679730733236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,4,1,balanced,0.87009596824646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,4,1,balanced,0.871514638264974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,4,1,balanced,0.808895985285441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,4,power_law_1.2,7.105197143554688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,4,1,balanced,0.8120693365732828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,4,power_law_1.2,0.6328256130218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,4,power_law_1.01,10.524537658691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,4,1,balanced,0.7829013665517172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,4,1,balanced,0.7902613480885824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,4,power_law_1.2,1.115436840057373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,4,1,balanced,0.8011519908905029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,4,1,balanced,0.8127466837565104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,4,power_law_1.2,1.15350399017334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,4,1,balanced,0.8391413688659668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,4,1,balanced,0.8560480276743571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,4,balanced,3.7779413859049478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,4,1,balanced,0.8830880324045817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,4,power_law_1.2,1.1830528259277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,4,1,balanced,0.09104532996813457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,4,1,balanced,0.15077867110570273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,4,1,balanced,1.1082293192545574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,4,1,balanced,0.2642773389816284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,4,power_law_1.2,1.2851072311401368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,4,1,balanced,1.156383991241455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,4,1,balanced,1.2591946919759114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,4,balanced,4.482629458109538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,4,power_law_1.2,1.3085056304931642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,4,power_law_1.2,9.273407745361329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,4,1,balanced,1.3526719411214192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,4,power_law_1.2,1.3700096130371093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,4,1,balanced,2.1419893900553384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,4,1,power_law_1.01,0.058233600854873654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,4,1,power_law_1.01,0.09422079920768738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,4,1,balanced,2.3386027018229165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,4,power_law_1.2,1.4482239723205566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,4,balanced,6.782768249511719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,4,1,power_law_1.01,0.15020159482955933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,4,power_law_1.01,15.191328430175782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,4,1,power_law_1.01,0.2519680023193359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,4,1,balanced,3.316015879313151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,4,1,power_law_1.01,0.33676159381866455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,4,1,balanced,3.7862987518310547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,4,1,power_law_1.01,0.5506752014160157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,4,balanced,8.59550412495931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,4,1,power_law_1.01,0.088646399974823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,4,1,power_law_1.01,0.14905600547790526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,4,power_law_1.2,13.557997131347657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,4,1,power_law_1.01,0.26635520458221434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,4,1,power_law_1.2,0.061452800035476686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,4,1,power_law_1.01,0.7060351848602295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,4,1,power_law_1.2,0.09466239809989929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,4,power_law_1.2,1.5604479789733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,4,1,power_law_1.01,0.7242879867553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,4,1,power_law_1.2,0.15304960012435914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,4,1,balanced,5.946362813313802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,4,1,balanced,0.5061333179473877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,4,1,power_law_1.01,0.68821120262146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,4,1,power_law_1.2,0.23598079681396483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,4,1,balanced,0.9422293504079183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,4,1,power_law_1.01,0.7080512046813965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,4,1,power_law_1.2,0.3187968015670776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,4,1,power_law_1.01,0.7062592029571533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,4,1,power_law_1.2,0.5367872238159179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,4,1,power_law_1.01,0.7219391822814941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,4,power_law_1.01,19.401242065429688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,4,1,balanced,7.013898849487305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,4,1,power_law_1.2,0.6958720207214355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,4,1,power_law_1.01,0.7361023902893067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,8,balanced,0.08872000376383464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,8,balanced,0.08903466661771138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,8,balanced,0.08773866295814514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,4,power_law_1.2,1.6661247253417968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,4,1,power_law_1.2,0.7238463878631591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,4,1,power_law_1.01,0.7581183910369873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,4,1,power_law_1.2,0.6847424030303955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,4,1,power_law_1.01,0.7708735942840577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,4,power_law_1.2,2.0066816329956056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,4,1,power_law_1.2,0.7065279960632325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,4,1,power_law_1.01,0.806003189086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,4,1,power_law_1.2,0.7023551940917969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,8,balanced,0.11296000083287557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,8,balanced,0.10922666390736897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,8,balanced,0.10922666390736897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,8,balanced,0.10965333382288615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,4,1,power_law_1.2,0.7239808082580567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,4,power_law_1.2,2.287251281738281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,8,balanced,0.1113759974638621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,8,balanced,0.11326400438944499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,8,balanced,0.0895253320535024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,8,balanced,0.13217600186665854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,8,balanced,0.08678399523099263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,8,balanced,0.13382933537165323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,8,balanced,0.08884800473848979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,8,balanced,0.1226026713848114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,4,1,power_law_1.2,0.7390655994415283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,8,balanced,0.12293333808581035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,8,balanced,0.1325279970963796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,8,balanced,0.13804266850153604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,8,balanced,0.12321600317955017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,8,balanced,0.14172800381978354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,4,1,power_law_1.2,0.7810944080352783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,8,balanced,0.14482667048772177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,8,balanced,0.15846932927767435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,4,power_law_1.2,2.9165056228637694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,8,balanced,0.17380799849828085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,8,balanced,0.18119466304779053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,8,balanced,0.1792853275934855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,4,1,power_law_1.2,0.8066431999206543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,8,balanced,0.21857066949208578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,8,balanced,0.24250133832295737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,4,1,power_law_1.01,0.8512895584106446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,8,balanced,0.3118026653925578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,8,balanced,0.3826719919840495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,4,1,power_law_1.2,0.8345664024353028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,4,1,power_law_1.2,0.08624640107154846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,8,balanced,0.5206933418909708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,4,1,power_law_1.01,1.0955072402954102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,8,balanced,0.19950399796168009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,8,balanced,0.6625653505325317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,4,1,power_law_1.2,0.15016319751739501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,4,power_law_1.2,3.502841567993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,4,1,power_law_1.2,0.8808383941650391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,4,1,power_law_1.01,0.4685567855834961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,8,balanced,0.9911253452301025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,4,1,power_law_1.01,1.1981439590454102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,4,1,power_law_1.2,0.2633280038833618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,8,balanced,1.2775786717732747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,4,1,power_law_1.2,1.1483712196350098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,4,1,power_law_1.01,0.6352320194244385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,4,1,power_law_1.01,1.3829504013061524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,8,balanced,1.9077119827270508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,4,1,balanced,1.428010622660319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,4,1,power_law_1.2,1.2347200393676758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,4,power_law_1.2,4.676204681396484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,4,1,balanced,1.4296746253967285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,4,1,power_law_1.01,1.5736767768859863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,8,balanced,2.562357266743978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,8,balanced,0.2010506590207418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,4,1,balanced,1.3923039436340332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,8,balanced,0.19075200955073038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,8,balanced,0.21338133017222086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,8,balanced,0.2175146738688151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,8,balanced,0.22562666734059653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,8,balanced,0.25331199169158936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,4,1,power_law_1.01,1.953183937072754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,8,balanced,0.2595253388086955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,8,balanced,0.259333332379659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,8,balanced,0.30210665861765545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,8,balanced,0.32259732484817505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,8,balanced,0.39346134662628174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,8,balanced,0.46721065044403076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,8,balanced,0.622544010480245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,4,1,power_law_1.01,2.3335168838500975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,4,power_law_1.2,6.222143936157226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,8,balanced,0.7805066903432211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,4,1,power_law_1.2,1.4314623832702638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,8,balanced,1.186794678370158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,4,1,balanced,1.395359992980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,4,1,power_law_1.2,1.6158464431762696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,8,balanced,1.5406452814737956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,4,1,balanced,1.4030399322509766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,4,1,power_law_1.01,3.1281856536865233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,8,balanced,2.323418617248535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,4,1,power_law_1.2,2.008198356628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,8,balanced,3.159701347351074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,4,1,power_law_1.2,2.4148672103881834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,4,1,power_law_1.01,3.911763381958008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,4,power_law_1.2,8.329779052734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,4,1,power_law_1.2,3.1629695892333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,4,1,power_law_1.01,5.526131057739258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,4,1,power_law_1.01,1.0651071548461915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,4,1,power_law_1.2,3.9740478515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,4,1,power_law_1.2,0.4137599945068359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,4,1,power_law_1.01,1.1552448272705078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,4,1,power_law_1.2,0.5664959907531738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,4,1,power_law_1.01,1.1759807586669921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,4,1,power_law_1.01,1.1830016136169434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,4,1,power_law_1.01,7.10680923461914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,8,balanced,0.12648000319798788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,4,power_law_1.2,10.74906234741211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,8,balanced,0.12597333391507468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,4,1,power_law_1.01,1.2202752113342286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,8,balanced,0.12770666678746542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,8,balanced,0.1276479959487915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,8,balanced,0.12772799531618753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,8,balanced,0.16146133343378702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,4,1,power_law_1.2,5.547020721435547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,8,balanced,0.14849600195884705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,8,balanced,0.1518826683362325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,8,balanced,0.220634659131368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,8,balanced,0.22059732675552368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,8,balanced,0.2257173260052999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,8,balanced,0.23629866043726602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,8,balanced,0.24060799678166708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,8,balanced,0.26150933901468915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,8,balanced,0.30134934186935425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,8,balanced,0.3136853377024333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,4,1,balanced,1.3762186368306477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,8,balanced,0.3263253370920817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,8,balanced,0.4033279816309611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,4,1,balanced,1.3843572934468586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,8,balanced,0.46719467639923096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,8,balanced,0.6016746759414673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,4,1,balanced,1.3998346328735352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,8,balanced,0.7491466999053955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,8,power_law_1.01,0.1096384048461914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,8,balanced,1.145466645558675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,4,1,power_law_1.2,7.176338958740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,8,power_law_1.01,0.10990719795227051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,8,power_law_1.01,0.11105279922485352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,8,balanced,1.4388373692830403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,8,power_law_1.01,0.11037440299987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,8,power_law_1.01,0.11089279651641845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,8,power_law_1.01,0.14191360473632814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,8,balanced,2.1228639284769693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,8,power_law_1.01,0.14402559995651246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,8,power_law_1.01,0.17383040189743043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,8,power_law_1.01,0.22086400985717775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,8,power_law_1.01,0.24268159866333008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,8,balanced,2.7963892618815103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,8,power_law_1.01,0.21639039516448974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,4,power_law_1.2,15.281721496582032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,8,power_law_1.01,0.08535040020942689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,8,balanced,4.183077176411946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,8,power_law_1.01,0.08524159789085388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,8,power_law_1.01,0.08488960266113281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,8,power_law_1.01,0.08535680174827576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,8,power_law_1.01,0.08610560297966004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,8,power_law_1.01,0.09719679951667785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,4,1,power_law_1.2,1.054412841796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,8,power_law_1.01,0.1371840000152588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,8,power_law_1.01,0.13803520202636718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,8,balanced,5.494848251342773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,8,power_law_1.01,0.15657600164413452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,4,1,power_law_1.2,1.1519552230834962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,8,power_law_1.01,0.16195839643478394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,8,power_law_1.01,0.13107839822769166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,4,1,power_law_1.2,1.1485312461853028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,4,1,power_law_1.2,1.2025856018066405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,4,1,power_law_1.01,1.2027135848999024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,4,1,power_law_1.2,1.215187168121338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,8,balanced,0.04091199984153112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,4,1,power_law_1.01,1.258835220336914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,8,balanced,0.041349334021409355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,8,balanced,0.041306667029857635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,8,balanced,0.04177066683769226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,8,balanced,0.04284266630808512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,8,power_law_1.01,0.12495360374450684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,8,power_law_1.01,0.12563199996948243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,4,1,power_law_1.01,1.3187775611877441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,8,power_law_1.01,0.1260543942451477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,4,1,balanced,1.443903923034668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,8,power_law_1.01,0.12727680206298828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,8,power_law_1.01,0.12784639596939087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,4,1,balanced,1.463941256205241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,8,power_law_1.01,0.16217600107192992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,8,power_law_1.01,0.18467199802398682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,4,1,balanced,1.6011892954508464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,8,power_law_1.01,0.16739840507507325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,8,power_law_1.01,0.21997439861297607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,4,1,balanced,1.8791467348734539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,8,power_law_1.01,0.23265280723571777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,4,power_law_1.2,20.614393615722655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,8,power_law_1.01,0.24972798824310302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,4,1,balanced,2.648944060007731
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,8,balanced,0.047877331574757896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,8,balanced,0.05566399792830149
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,8,balanced,0.05594133337338766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,8,balanced,0.1745599905649821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,8,balanced,0.1665279964605967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,8,balanced,0.1520693302154541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,8,balanced,0.15271466970443726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,8,power_law_1.01,0.28988161087036135
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,8,balanced,0.14527466893196106
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,8,balanced,0.12483732899030049
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,8,balanced,0.11716799934705098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,8,balanced,0.07005333403746287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,8,balanced,0.07065600156784058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,8,power_law_1.01,0.29701759815216067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,8,power_law_1.01,0.3314111948013306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,8,power_law_1.01,0.3501375913619995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,8,power_law_1.01,0.2326591968536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,8,power_law_1.01,0.38782079219818116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,8,power_law_1.01,0.23890559673309325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,8,power_law_1.01,0.2622143983840942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,8,balanced,0.08550399541854858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,8,power_law_1.01,0.4644480228424072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,8,balanced,0.09249066313107808
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,8,balanced,0.11211733023325603
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,8,balanced,0.1311893363793691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,8,balanced,0.18028799692789713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,8,balanced,0.2198986609776815
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,8,balanced,0.31165866057078045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,8,balanced,0.40780266125996906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,4,1,power_law_1.2,1.2411904335021973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,8,balanced,0.5911733309427897
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,8,balanced,0.7732799847920736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,4,1,power_law_1.2,1.3007295608520508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,8,power_law_1.01,0.14833920001983641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,4,1,power_law_1.01,1.3993151664733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,8,power_law_1.01,0.2877311944961548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,8,power_law_1.01,0.15421439409255983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,4,1,power_law_1.2,1.369484806060791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,8,power_law_1.01,0.29691519737243655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,8,power_law_1.01,0.16840319633483886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,4,1,power_law_1.01,1.491648006439209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,8,power_law_1.01,0.3469183921813965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,4,1,balanced,2.5486987431844077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,8,power_law_1.01,0.18892159461975097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,4,1,power_law_1.01,1.6025407791137696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,8,power_law_1.01,0.6888256072998047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,8,power_law_1.01,0.19816960096359254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,4,1,balanced,3.9888480504353843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,8,power_law_1.2,0.10730240345001221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,8,power_law_1.01,0.3815936088562012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,8,power_law_1.2,0.1083840012550354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,4,1,power_law_1.01,1.8668096542358399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,8,power_law_1.2,0.10878080129623413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,8,power_law_1.01,0.43800959587097166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,8,power_law_1.2,0.1102720022201538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,8,power_law_1.01,0.8346176147460938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,8,power_law_1.2,0.10991359949111938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,8,power_law_1.2,0.1486143946647644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,8,power_law_1.2,0.16292480230331421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,4,1,balanced,4.757114728291829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,4,1,power_law_1.01,2.415807914733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,8,power_law_1.2,0.1692031979560852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,8,power_law_1.01,0.7173056125640869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,8,power_law_1.2,0.22690560817718505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,8,power_law_1.2,0.24143359661102295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,8,power_law_1.2,0.2183295965194702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,8,power_law_1.01,0.21497600078582763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,8,power_law_1.2,0.23257598876953126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,8,power_law_1.2,0.2389888048171997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,4,1,balanced,7.016266504923503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,8,power_law_1.01,0.24776959419250488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,8,power_law_1.2,0.26150400638580323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,8,power_law_1.01,0.8532671928405762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,8,power_law_1.2,0.2853440046310425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,8,power_law_1.01,0.2996992111206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,8,power_law_1.01,1.2447936058044433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,8,power_law_1.2,0.31617920398712157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,4,1,balanced,8.345930735270182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,8,power_law_1.2,0.32636799812316897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,8,power_law_1.01,0.44586877822875975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,8,power_law_1.2,0.39107840061187743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,8,power_law_1.01,1.554975986480713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,4,1,power_law_1.2,1.4401663780212401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,4,1,power_law_1.2,1.541708755493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,4,1,balanced,13.1724001566569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,8,power_law_1.01,0.5368256092071533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,4,1,power_law_1.2,1.6453119277954102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,8,power_law_1.01,0.041017600893974306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,8,power_law_1.01,0.041382399201393125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,8,power_law_1.01,1.318502426147461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,4,1,power_law_1.01,2.7285823822021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,8,power_law_1.01,0.04087679982185364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,4,1,power_law_1.2,1.8989887237548828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,8,power_law_1.01,0.04099839925765991
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,8,power_law_1.01,0.04154880046844482
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,8,power_law_1.01,0.04508160054683685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,8,power_law_1.01,0.053311997652053834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,8,power_law_1.01,0.05253120064735413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,4,1,power_law_1.2,2.4122175216674804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,4,1,power_law_1.01,3.744998550415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,8,power_law_1.01,0.10599039793014527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,8,power_law_1.01,0.11670399904251098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,8,power_law_1.01,0.058169597387313844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,8,power_law_1.01,0.07747200131416321
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,8,power_law_1.01,0.10279040336608887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,8,power_law_1.2,0.4798783779144287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,8,power_law_1.01,0.07118719816207886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,4,1,balanced,16.68433125813802
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,8,power_law_1.01,0.1289791941642761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,4,1,power_law_1.01,4.729324722290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,8,power_law_1.01,2.4061248779296873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,8,power_law_1.01,0.7576320171356201
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,8,power_law_1.01,0.08228480219841003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,8,power_law_1.01,0.0923200011253357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,4,1,power_law_1.01,6.624800109863282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,8,power_law_1.01,0.12321280241012574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,8,power_law_1.01,1.01278076171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,8,power_law_1.01,0.1454848051071167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,8,power_law_1.2,0.7234687805175781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,8,power_law_1.01,0.1951359987258911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,4,1,power_law_1.01,8.780793762207031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,4,1,balanced,24.774373372395832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,8,power_law_1.2,0.8784383773803711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,4,1,power_law_1.2,2.7566911697387697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,8,power_law_1.01,1.766092872619629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,8,power_law_1.2,1.2721471786499023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,4,1,power_law_1.2,3.7570110321044923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,8,power_law_1.01,2.978348731994629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,4,1,power_law_1.01,12.985670471191407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,4,1,power_law_1.2,4.755923080444336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,8,power_law_1.01,0.24424960613250732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,8,power_law_1.01,1.5760255813598634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,4,1,power_law_1.2,6.611775970458984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,8,power_law_1.01,2.483225631713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,4,1,balanced,32.747510274251304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,4,1,power_law_1.01,16.76817932128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,4,1,power_law_1.2,8.808761596679688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,8,power_law_1.01,4.115705490112305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,4,1,power_law_1.2,12.970477294921874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,8,power_law_1.01,0.3441792011260986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,8,power_law_1.2,0.08579840064048767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,8,power_law_1.2,0.08588799834251404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,8,power_law_1.2,0.0864192008972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,8,power_law_1.01,5.674822235107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,8,power_law_1.2,1.7026752471923827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,4,1,power_law_1.01,25.0413818359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,8,power_law_1.01,2.102457618713379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,8,power_law_1.01,3.2635711669921874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,4,1,power_law_1.2,16.91080322265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,8,power_law_1.01,0.46816000938415525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,8,power_law_1.2,0.08633599877357483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,8,power_law_1.2,0.08661119937896729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,8,power_law_1.2,0.10081280469894409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,4,1,power_law_1.01,33.03541870117188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,4,1,power_law_1.2,24.675494384765624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,8,power_law_1.2,0.13359999656677246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,8,power_law_1.2,2.6647552490234374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,8,power_law_1.2,0.1383936047554016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,8,power_law_1.01,0.6544384002685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,8,power_law_1.01,0.777952003479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,8,power_law_1.01,5.949798583984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,8,power_law_1.2,0.16382720470428466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,8,power_law_1.2,0.16504960060119628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,4,1,power_law_1.2,32.82808227539063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,8,power_law_1.01,9.22476806640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,8,power_law_1.01,3.3046207427978516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,8,power_law_1.2,0.13259520530700683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,8,power_law_1.01,7.3360130310058596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,8,1,balanced,0.03565333286921183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,8,1,balanced,0.05618133147557577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,8,1,balanced,0.09696533282597859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,8,1,balanced,0.09717333316802979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,8,1,balanced,0.09755733609199524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,8,power_law_1.2,3.637638473510742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,8,1,balanced,0.09882666667302449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,8,1,balanced,0.08338666955629985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,8,1,balanced,0.08378133177757263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,8,1,balanced,0.09655466675758362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,8,1,balanced,0.09874666730562846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,8,1,balanced,0.09533333778381348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,8,1,balanced,0.09705066680908203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,8,1,balanced,0.09815466403961182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,8,1,balanced,0.10285333792368571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,8,1,balanced,0.14418133099873862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,8,1,balanced,0.15076800187428793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,8,1,balanced,0.13966932892799377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,8,1,balanced,0.15636266271273294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,8,1,balanced,0.16248533129692078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,8,1,balanced,0.2222879926363627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,8,1,balanced,0.27755733331044513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,8,1,balanced,0.3731360038121541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,8,1,balanced,0.4779306650161743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,8,1,balanced,0.6878133614857992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,8,1,balanced,0.8676640192667643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,8,1,balanced,1.2945120334625244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,8,1,balanced,1.7163999875386555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,8,1,balanced,0.031248000760873158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,8,1,balanced,0.043162668744723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,8,1,balanced,0.05997333427270254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,8,1,balanced,0.059157331784566246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,8,1,balanced,0.05983466903368632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,8,1,balanced,0.06328000128269196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,8,1,balanced,0.07523199915885925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,8,1,balanced,0.07656533519426982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,8,1,balanced,0.07498133182525635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,8,power_law_1.2,0.03928320109844208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,8,1,balanced,0.07840533554553986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,8,1,balanced,0.0697813332080841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,8,1,balanced,0.07223466535409291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,8,1,balanced,0.07092800239721934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,8,1,balanced,0.07630399862925212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,8,1,balanced,0.10115733742713928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,8,1,balanced,0.1025279959042867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,8,1,balanced,0.09405333797136943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,8,1,balanced,0.11155200004577637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,8,1,balanced,0.1145919958750407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,8,1,balanced,0.15680533647537231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,8,1,balanced,0.19286400079727173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,8,1,balanced,0.25627734263737995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,8,1,balanced,0.32763733466466266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,8,1,balanced,0.45553600788116455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,8,1,balanced,0.5911999940872192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,8,1,balanced,0.852837324142456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,8,power_law_1.2,0.14657919406890868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,8,power_law_1.2,0.03903360068798065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,8,1,balanced,1.1270559628804524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,8,power_law_1.2,0.15599360466003417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,8,power_law_1.2,0.039027199149131775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,8,1,balanced,0.041893333196640015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,8,1,balanced,0.05453866720199585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,8,power_law_1.2,0.040652799606323245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,8,1,balanced,0.06434666613737743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,8,1,balanced,0.06588799754778545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,8,1,balanced,0.06663999954859416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,8,power_law_1.2,0.03947519958019256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,8,1,balanced,0.07036800185839336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,8,1,balanced,0.07243200143178304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,8,1,balanced,0.07229866584142049
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,8,power_law_1.2,0.04471679925918579
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,8,power_law_1.01,1.2926464080810547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,8,1,balanced,0.08860266208648682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,8,1,balanced,0.08943466345469157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,8,1,balanced,0.0935093363126119
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,8,power_law_1.2,0.051974397897720334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,8,1,balanced,0.09648533662160237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,8,1,balanced,0.09611200292905171
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,8,power_law_1.2,0.052095997333526614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,8,1,balanced,0.10319466392199199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,8,1,balanced,0.1548426647981008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,8,power_law_1.2,0.09842560291290284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,8,power_law_1.2,0.0881600022315979
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,8,power_law_1.2,0.05873280167579651
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,8,power_law_1.2,0.10362880229949951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,8,power_law_1.2,0.17326079607009887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,8,power_law_1.2,0.12702080011367797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,8,power_law_1.2,0.13475199937820434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,8,power_law_1.2,0.12716799974441528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,8,power_law_1.2,0.18880000114440917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,8,power_law_1.2,0.12821760177612304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,8,power_law_1.2,0.07239680290222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,8,power_law_1.2,0.12821120023727417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,8,1,balanced,0.15582399566968283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,8,power_law_1.2,0.20032000541687012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,8,1,balanced,0.15455999970436096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,8,1,balanced,0.21184533834457397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,8,1,balanced,0.26446400086085003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,8,1,balanced,0.3527146577835083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,8,1,balanced,0.4496693213780721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,8,1,balanced,0.6311893463134766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,8,power_law_1.2,0.21822719573974608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,8,1,balanced,0.8092373212178549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,8,1,balanced,1.1665120124816895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,8,1,balanced,1.5329012870788574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,8,power_law_1.2,0.12946560382843017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,8,power_law_1.2,0.16410239934921264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,8,power_law_1.2,0.16363519430160522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,8,power_law_1.2,0.09820160269737244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,8,1,balanced,2.2047252655029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,8,power_law_1.2,0.16528639793395997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,8,power_law_1.2,0.23117439746856688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,8,power_law_1.2,0.08408960103988647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,8,power_law_1.2,0.24408960342407227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,8,1,balanced,2.9050025939941406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,8,power_law_1.2,0.25022718906402586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,8,power_law_1.2,0.28504319190979005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,8,power_law_1.2,0.2907776117324829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,8,power_law_1.2,0.31485440731048586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,8,power_law_1.2,0.09400960206985473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,8,power_law_1.2,0.35676798820495603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,8,power_law_1.2,0.4027775764465332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,8,power_law_1.2,0.4486720085144043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,8,power_law_1.2,0.1235200047492981
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,8,1,balanced,0.029253333806991577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,8,1,balanced,0.029311999678611755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,8,1,balanced,0.039120001097520195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,8,1,balanced,0.03937600056330363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,8,1,balanced,0.039493332306543984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,8,1,balanced,0.03982399900754293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,8,1,balanced,0.04769066472848257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,8,1,balanced,0.048767998814582825
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,8,1,balanced,0.051221330960591636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,8,1,balanced,0.0516533354918162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,8,1,balanced,0.05086400111516317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,8,1,balanced,0.05477866530418396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,8,1,balanced,0.05558399856090546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,8,1,balanced,0.05615466833114624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,8,1,balanced,0.08140799899895985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,8,1,balanced,0.07155199845631917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,8,1,balanced,0.06429333488146464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,8,power_law_1.2,0.6297344207763672
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,8,1,balanced,0.1172106663386027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,8,1,balanced,0.08596799770991008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,8,1,balanced,0.10065600275993347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,8,1,balanced,0.1237333317597707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,8,1,balanced,0.16365333398183188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,8,1,balanced,0.20636266469955444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,8,1,balanced,0.2929919958114624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,8,1,balanced,0.37834668159484863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,8,power_law_1.2,0.8401727676391602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,8,power_law_1.2,0.15294719934463502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,8,power_law_1.2,1.1539008140563964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,8,1,balanced,0.5493653217951456
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,8,1,balanced,0.7147839864095052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,8,1,power_law_1.01,0.03375999927520752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,8,1,power_law_1.01,0.04796159863471985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,8,1,power_law_1.01,0.058220797777175905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,8,1,power_law_1.01,0.08490239977836608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,8,1,power_law_1.01,0.08543360233306885
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,8,power_law_1.2,0.20576000213623047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,8,1,power_law_1.01,0.09274240136146546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,8,1,power_law_1.01,0.08188160061836243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,8,1,power_law_1.01,0.0827455997467041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,8,power_law_1.2,0.25662078857421877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,8,1,power_law_1.01,0.09854080080986023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,8,power_law_1.2,0.24389119148254396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,8,1,power_law_1.01,0.10107519626617431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,8,1,power_law_1.01,0.09422720074653626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,8,1,power_law_1.01,0.09761919975280761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,8,1,power_law_1.01,0.09926400184631348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,8,1,power_law_1.01,0.10291839838027954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,8,1,power_law_1.01,0.12210559844970703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,8,1,power_law_1.01,0.1475711941719055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,8,1,power_law_1.01,0.1428928017616272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,8,power_law_1.2,0.3126976013183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,8,1,power_law_1.01,0.15723520517349243
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,8,power_law_1.01,1.7256704330444337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,8,1,power_law_1.01,0.0324288010597229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,8,1,power_law_1.01,0.03831680119037628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,8,1,power_law_1.01,0.04437119960784912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,8,1,power_law_1.01,0.05330560207366943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,8,1,power_law_1.01,0.05413119792938233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,8,1,power_law_1.01,0.1860479950904846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,8,1,power_law_1.01,0.056460797786712646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,8,1,power_law_1.01,0.07036799788475037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,8,1,power_law_1.01,0.07466880083084107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,8,1,power_law_1.01,0.08471680283546448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,8,power_law_1.2,0.3605247974395752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,8,power_law_1.2,1.5725248336791993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,8,1,power_law_1.01,0.2387712001800537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,8,power_law_1.2,0.44298877716064455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,8,1,power_law_1.01,0.08602240085601806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,8,1,power_law_1.01,0.06602240204811097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,8,1,power_law_1.01,0.06967679858207702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,8,1,power_law_1.01,0.07259520292282104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,8,1,power_law_1.01,0.07633280158042907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,8,1,power_law_1.01,0.089055997133255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,8,1,power_law_1.01,0.10246399641036988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,8,1,power_law_1.01,0.09541760087013244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,8,1,power_law_1.01,0.10855040550231934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,8,1,power_law_1.01,0.12743680477142333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,8,power_law_1.2,0.5444159984588623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,8,power_law_1.2,2.4377792358398436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,8,1,power_law_1.01,0.28721280097961427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,8,1,power_law_1.01,0.16355839967727662
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,8,power_law_1.2,0.469158411026001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,8,1,power_law_1.01,0.20317440032958983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,8,power_law_1.01,12.129574584960938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,8,1,power_law_1.01,0.0463808000087738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,8,1,power_law_1.01,0.051475197076797485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,8,1,power_law_1.01,0.05441280007362366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,8,power_law_1.2,5.6821247100830075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,8,1,power_law_1.01,0.3827392101287842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,8,power_law_1.01,4.490623855590821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,8,1,power_law_1.01,0.2619904041290283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,8,power_law_1.2,0.8316864013671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,8,1,power_law_1.01,0.3294975996017456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,8,1,power_law_1.01,0.06380800008773804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,8,1,power_law_1.01,0.06334720253944397
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,8,1,power_law_1.01,0.02484479993581772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,8,1,power_law_1.01,0.025990399718284606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,8,1,power_law_1.01,0.029497599601745604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,8,1,power_law_1.01,0.03378559947013855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,8,1,power_law_1.01,0.03857919871807099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,8,1,power_law_1.01,0.041555199027061465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,8,1,power_law_1.01,0.045926401019096376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,8,1,power_law_1.01,0.04710400104522705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,8,1,power_law_1.01,0.04639999866485596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,8,1,power_law_1.01,0.04880639910697937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,8,1,power_law_1.01,0.05137280225753784
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,8,1,power_law_1.01,0.05230720043182373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,8,1,power_law_1.01,0.054502397775650024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,8,1,power_law_1.01,0.05866879820823669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,8,power_law_1.2,2.9378047943115235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,8,1,power_law_1.01,0.4942336082458496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,8,1,power_law_1.01,0.0607807993888855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,8,1,power_law_1.01,0.06323199868202209
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,8,1,power_law_1.01,0.06807039976119995
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,8,1,power_law_1.01,0.08165760040283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,8,1,power_law_1.01,0.06873599886894226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,8,1,power_law_1.01,0.06936320066452026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,8,1,power_law_1.01,0.07489920258522034
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,8,1,power_law_1.01,0.08944000005722046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,8,1,power_law_1.01,0.10649600028991699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,8,1,power_law_1.01,0.1241536021232605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,8,1,power_law_1.01,0.6963647842407227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,8,1,power_law_1.01,0.17077759504318238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,8,1,power_law_1.01,0.4640063762664795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,8,power_law_1.2,1.1094592094421387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,8,1,power_law_1.01,0.08714879751205444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,8,power_law_1.2,0.6961919784545898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,8,1,power_law_1.01,0.09106559753417968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,8,1,power_law_1.01,0.09856640100479126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,8,1,power_law_1.01,0.11320960521697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,8,1,power_law_1.01,0.12780159711837769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,8,1,power_law_1.01,0.14542720317840577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,8,1,power_law_1.01,0.15056639909744263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,8,1,power_law_1.01,0.15550719499588012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,8,1,power_law_1.01,0.17299840450286866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,8,1,power_law_1.01,0.2085184097290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,8,power_law_1.2,4.637113571166992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,8,1,power_law_1.01,0.22035839557647705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,8,1,power_law_1.01,0.27231359481811523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,8,1,power_law_1.01,0.5946752071380615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,8,1,power_law_1.01,0.3542912006378174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,8,1,power_law_1.01,0.2957312107086182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,8,1,power_law_1.01,0.45770878791809083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,8,1,power_law_1.01,0.9106623649597168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,8,power_law_1.2,7.630316925048828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,8,power_law_1.2,1.6241727828979493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,8,1,power_law_1.2,0.034745600819587705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,8,1,power_law_1.2,0.04214400053024292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,8,1,power_law_1.01,0.635481595993042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,8,1,power_law_1.2,0.05234559774398804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,8,1,power_law_1.2,0.08297600150108338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,8,1,power_law_1.2,0.08620799779891967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,8,1,power_law_1.2,0.08707200288772583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,8,1,power_law_1.2,0.0808896005153656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,8,1,power_law_1.01,0.8509951591491699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,8,1,power_law_1.2,0.08539519906044006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,8,1,power_law_1.2,0.09935359954833985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,8,1,power_law_1.2,0.09984639883041382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,8,1,power_law_1.2,0.09338240027427673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,8,1,power_law_1.2,0.0949504017829895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,8,1,power_law_1.2,0.09828479886054993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,8,1,power_law_1.2,0.10843520164489746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,8,1,power_law_1.2,0.11653759479522705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,8,1,power_law_1.2,0.14774399995803833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,8,1,power_law_1.01,0.8056960105895996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,8,1,power_law_1.2,0.14296319484710693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,8,1,power_law_1.01,0.38728320598602295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,8,power_law_1.2,6.158163070678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,8,power_law_1.2,0.902956771850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,8,1,power_law_1.01,1.1559359550476074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,8,1,power_law_1.2,0.157151997089386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,8,1,power_law_1.2,0.18826240301132202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,8,1,power_law_1.2,0.2384511947631836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,8,1,power_law_1.01,1.3184191703796386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,8,power_law_1.2,2.1203903198242187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,8,1,power_law_1.2,0.28465919494628905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,8,1,power_law_1.2,0.39604480266571046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,8,1,power_law_1.01,1.4852928161621093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,8,1,power_law_1.01,1.1239423751831055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,8,1,power_law_1.2,0.03319680094718933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,8,1,power_law_1.2,0.03882879912853241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,8,1,power_law_1.2,0.04343680143356323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,8,1,power_law_1.2,0.052928000688552856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,8,1,power_law_1.2,0.05518720149993896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,8,1,power_law_1.2,0.05663999915122986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,8,1,power_law_1.2,0.07004160284996033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,8,1,power_law_1.2,0.0721343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,8,1,power_law_1.2,0.08319360017776489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,8,1,power_law_1.2,0.0866047978401184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,8,1,power_law_1.2,0.06633599996566772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,8,1,power_law_1.2,0.0692799985408783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,8,1,power_law_1.2,0.07141759991645813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,8,1,power_law_1.2,0.07693439722061157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,8,1,power_law_1.01,0.5530176162719727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,8,1,power_law_1.2,0.08640000224113464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,8,1,power_law_1.2,0.10309760570526123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,8,1,power_law_1.2,0.0942463994026184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,8,1,power_law_1.2,0.10981119871139526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,8,1,power_law_1.2,0.495743989944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,8,1,power_law_1.2,0.1238144040107727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,8,1,power_law_1.2,0.6947775840759277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,8,1,power_law_1.2,0.16601599454879762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,8,1,power_law_1.01,1.7264064788818358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,8,1,power_law_1.2,0.0495743989944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,8,1,power_law_1.2,0.05374720096588135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,8,1,power_law_1.2,0.0553600013256073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,8,1,power_law_1.2,0.06723200082778931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,8,1,power_law_1.2,0.06529279947280883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,8,1,power_law_1.2,0.07006080150604248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,8,1,power_law_1.2,0.07125120162963867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,8,power_law_1.2,9.14554214477539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,8,1,power_law_1.2,0.0769919991493225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,8,1,power_law_1.2,0.08737279772758484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,8,1,power_law_1.2,0.09199360013008118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,8,1,power_law_1.2,0.1958400011062622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,8,1,power_law_1.2,0.09605119824409485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,8,1,power_law_1.2,0.11727360486984253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,8,1,power_law_1.2,0.14016640186309814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,8,1,power_law_1.2,0.145632004737854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,8,1,power_law_1.2,0.1516095995903015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,8,1,power_law_1.2,0.15603840351104736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,8,1,power_law_1.2,0.184006404876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,8,1,power_law_1.2,0.2220479965209961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,8,1,power_law_1.01,2.1792768478393554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,8,1,power_law_1.2,0.26995201110839845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,8,1,power_law_1.2,0.26584320068359374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,8,1,power_law_1.2,0.3635967969894409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,8,1,power_law_1.2,0.33422720432281494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,8,1,power_law_1.2,0.45995521545410156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,8,power_law_1.2,1.3104000091552734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,8,power_law_1.2,3.19051513671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,8,1,power_law_1.01,0.7306687831878662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,8,1,power_law_1.2,0.4652095794677734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,8,1,power_law_1.2,0.6288064002990723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,8,1,power_law_1.2,0.02475520074367523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,8,1,power_law_1.2,0.026265600323677064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,8,1,power_law_1.2,0.029177600145339967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,8,1,power_law_1.2,0.03362559974193573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,8,1,power_law_1.2,0.03735679984092712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,8,1,power_law_1.2,0.9069760322570801
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,8,1,power_law_1.2,0.04193280041217804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,8,1,power_law_1.2,0.04629760086536407
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,8,1,power_law_1.2,0.048416000604629514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,8,1,power_law_1.2,0.04678399860858917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,8,1,power_law_1.2,0.04792959988117218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,8,1,power_law_1.2,0.05175679922103882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,8,1,power_law_1.2,0.05233280062675476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,8,1,power_law_1.2,0.05455999970436096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,8,1,power_law_1.2,0.06000000238418579
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,8,1,power_law_1.2,0.061945599317550656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,8,1,power_law_1.2,0.06282240152359009
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,8,1,power_law_1.2,0.06780160069465638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,8,1,power_law_1.2,0.8093952178955078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,8,1,power_law_1.2,0.07876480221748353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,8,1,power_law_1.2,0.0902783989906311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,8,power_law_1.2,11.845753479003907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,8,balanced,0.15154666701952615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,8,balanced,0.1490613321463267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,8,balanced,0.14828266700108847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,8,balanced,0.15380799770355225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,8,balanced,0.15268799662590027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,8,balanced,0.15340800086657205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,8,balanced,0.167903999487559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,8,balanced,0.16987200578053793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,8,balanced,0.254528005917867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,8,balanced,0.2559093236923218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,8,1,power_law_1.2,0.11398400068283081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,8,balanced,0.23919467131296793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,8,1,power_law_1.2,0.12446719408035278
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,8,1,power_law_1.2,0.17091200351715088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,8,1,power_law_1.2,1.1453696250915528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,8,balanced,0.2667466600735982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,8,balanced,0.2677653431892395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,8,balanced,0.28275734186172485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,8,balanced,0.32366400957107544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,8,balanced,0.3298506736755371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,8,balanced,0.32918934027353924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,8,1,power_law_1.01,2.8610687255859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,8,balanced,0.3789066473642985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,8,balanced,0.11230400204658508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,8,balanced,0.11131200194358826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,8,balanced,0.10973333319028218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,8,balanced,0.1129813293615977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,8,balanced,0.11266666650772095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,8,balanced,0.11375466982523601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,8,balanced,0.15874666968981424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,8,balanced,0.15892266233762106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,8,balanced,0.17906665802001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,8,balanced,0.17887999614079794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,8,balanced,0.1482080022493998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,8,balanced,0.17364799976348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,8,1,power_law_1.2,0.5934976100921631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,8,balanced,0.4030933380126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,8,balanced,0.49798933664957684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,8,balanced,0.5994133154551188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,8,balanced,0.8124639987945557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,8,balanced,1.1318346659342449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,8,balanced,0.17908267180124918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,8,balanced,0.19197867314020792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,8,balanced,0.21633599201838175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,8,balanced,0.22842133045196533
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,8,power_law_1.2,1.7381824493408202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,8,balanced,1.5914559364318848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,8,balanced,0.22714666525522867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,8,balanced,0.2762506604194641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,8,balanced,0.30608532826105755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,8,balanced,0.39182400703430176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,8,balanced,2.134314695994059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,8,balanced,0.48163731892903644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,8,balanced,0.666538675626119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,8,balanced,0.8969759941101074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,8,balanced,3.320672035217285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,8,balanced,1.278106689453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,8,balanced,1.6879253387451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,8,balanced,4.503135999043782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,8,balanced,2.5421652793884277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,8,1,power_law_1.2,0.8606592178344726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,8,balanced,3.4334239959716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,8,power_law_1.01,0.1509951949119568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,8,power_law_1.01,0.1497215986251831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,8,power_law_1.01,0.15145599842071533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,8,power_law_1.01,0.15012480020523072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,8,power_law_1.01,0.15141760110855101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,8,power_law_1.01,0.18042880296707153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,8,power_law_1.01,0.20758399963378907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,8,power_law_1.01,0.23463680744171142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,8,power_law_1.01,0.2803584098815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,8,power_law_1.01,0.319648003578186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,8,power_law_1.01,0.2857280015945435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,8,power_law_1.01,0.2996223926544189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,8,power_law_1.01,0.30592639446258546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,8,1,power_law_1.2,0.209881591796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,8,power_law_1.01,0.3315327882766724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,8,power_law_1.01,0.375545597076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,8,power_law_1.01,0.4020224094390869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,8,power_law_1.01,0.4638976097106934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,8,power_law_1.01,0.5388095855712891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,8,power_law_1.01,0.6984831809997558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,8,1,power_law_1.2,0.29832959175109863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,8,power_law_1.01,1.0158080101013183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,8,1,power_law_1.2,1.4971712112426758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,8,1,power_law_1.2,1.3114496231079102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,8,power_law_1.01,1.3217151641845704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,8,power_law_1.2,4.522880172729492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,8,power_law_1.01,0.11896320581436157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,8,power_law_1.01,0.11952639818191528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,8,power_law_1.01,0.11431039571762085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,8,balanced,0.05727999905745188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,8,balanced,0.056789333621660866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,8,power_law_1.01,0.11455359458923339
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,8,balanced,0.056927998860677086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,8,balanced,0.056218668818473816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,8,power_law_1.01,0.11426559686660767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,8,balanced,0.057909334699312844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,8,balanced,0.06102933486302694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,8,power_law_1.01,0.1273151993751526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,8,power_law_1.01,0.17051520347595214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,8,power_law_1.01,0.18929920196533204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,8,power_law_1.01,0.20805120468139648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,8,power_law_1.01,0.24058880805969238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,8,power_law_1.01,2.068511962890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,8,power_law_1.01,0.172326397895813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,8,power_law_1.01,0.18862080574035645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,8,power_law_1.01,0.1999935984611511
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,8,balanced,0.07183466851711273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,8,power_law_1.01,0.21960959434509278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,8,power_law_1.01,0.24507520198822022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,8,power_law_1.01,0.27153279781341555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,8,power_law_1.01,0.2760256052017212
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,8,balanced,0.08717866738637288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,8,balanced,0.21409066518147787
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,8,balanced,0.1913706660270691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,8,balanced,0.19114667177200317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,8,balanced,0.18463999032974243
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,8,balanced,0.17779199282328287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,8,balanced,0.1514240006605784
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,8,balanced,0.1381173332532247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,8,balanced,0.08053866525491078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,8,power_law_1.01,0.35871999263763427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,8,balanced,0.08590933680534363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,8,balanced,0.10916266838709514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,8,balanced,0.1156160036722819
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,8,balanced,0.14597866932551065
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,8,balanced,0.1876479983329773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,8,balanced,0.2507413427035014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,8,balanced,0.32453866799672443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,8,power_law_1.01,2.6798015594482423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,8,balanced,0.4715840021769206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,8,balanced,0.619269331296285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,8,balanced,0.9252106348673502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,8,balanced,1.2330719629923503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,8,power_law_1.01,0.49300479888916016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,8,power_law_1.01,0.15752960443496705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,8,power_law_1.01,0.15790079832077025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,8,power_law_1.01,0.1572543978691101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,8,power_law_1.01,0.1578112006187439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,8,power_law_1.01,0.161190402507782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,8,power_law_1.01,0.17913600206375122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,8,power_law_1.01,0.20776960849761963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,8,power_law_1.01,0.24138879776000977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,8,power_law_1.01,0.2724096059799194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,8,power_law_1.01,0.2827136039733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,8,power_law_1.01,0.31370880603790285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,8,power_law_1.01,0.35272960662841796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,8,power_law_1.01,0.3773888111114502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,8,power_law_1.01,0.6394239902496338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,8,power_law_1.01,0.4162303924560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,8,power_law_1.01,0.49102082252502444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,8,power_law_1.01,0.5071616172790527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,8,power_law_1.01,0.7344575881958008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,8,power_law_1.01,0.8216832160949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,8,power_law_1.01,0.9331904411315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,8,power_law_1.01,3.8617984771728517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,8,power_law_1.01,1.1658432006835937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,8,power_law_1.01,1.7463039398193358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,8,power_law_1.01,1.2754048347473144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,8,1,power_law_1.2,2.1795328140258787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,8,power_law_1.01,2.472652816772461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,8,1,power_law_1.2,1.1338560104370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,8,power_law_1.01,3.3810943603515624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,8,1,power_law_1.2,0.39375360012054444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,8,balanced,0.1585599978764852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,8,balanced,0.1576479971408844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,8,balanced,0.15890666842460632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,8,balanced,0.1594986617565155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,8,balanced,0.15981333454449972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,8,balanced,0.17767467101415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,8,balanced,0.18837867180506387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,8,balanced,0.19072532653808594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,8,balanced,0.2667520046234131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,8,balanced,0.27062400182088214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,8,balanced,0.2748426596323649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,8,power_law_1.01,1.5530943870544434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,8,balanced,0.29100799560546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,8,balanced,0.2950719992319743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,8,balanced,0.3166079918543498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,8,balanced,0.37380266189575195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,8,balanced,0.39189334710439044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,8,balanced,0.4076160192489624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,8,balanced,0.5102186600367228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,8,balanced,0.5984586477279663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,8,power_law_1.01,5.322259140014649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,8,balanced,0.7873653570810953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,8,balanced,1.0927093029022217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,8,power_law_1.01,4.867641448974609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,8,balanced,1.4932799339294434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,8,balanced,1.9954454104105632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,8,balanced,2.918992042541504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,8,balanced,3.8844372431437173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,8,power_law_1.01,2.410534477233887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,8,balanced,5.764682769775391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,8,balanced,7.6196746826171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,8,power_law_1.2,0.1511296033859253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,8,power_law_1.2,0.151910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,8,power_law_1.2,0.15170559883117676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,8,power_law_1.2,0.1498944044113159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,8,power_law_1.2,0.1510975956916809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,8,power_law_1.2,0.19374719858169556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,8,power_law_1.2,0.18176000118255614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,8,power_law_1.01,6.582733154296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,8,power_law_1.2,0.23790080547332765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,8,power_law_1.2,0.28302080631256105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,8,power_law_1.2,0.33761279582977294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,8,power_law_1.2,0.28627839088439944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,8,power_law_1.01,0.053887999057769774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,8,power_law_1.01,0.05419520139694214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,8,power_law_1.01,0.054118400812149046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,8,power_law_1.01,0.05367040038108826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,8,power_law_1.01,0.054585599899291994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,8,power_law_1.01,0.05825279951095581
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,8,power_law_1.01,0.06627200245857238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,8,power_law_1.01,0.06726400256156921
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,8,power_law_1.01,0.12495360374450684
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,8,power_law_1.01,0.12872960567474365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,8,power_law_1.01,0.07214080095291138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,8,power_law_1.01,0.10205440521240235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,8,power_law_1.01,0.1331264019012451
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,8,power_law_1.01,0.09342079758644103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,8,power_law_1.01,0.16499840021133422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,8,power_law_1.01,0.10902400016784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,8,power_law_1.01,0.1467136025428772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,8,power_law_1.01,0.17660800218582154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,8,power_law_1.01,0.22231678962707518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,8,power_law_1.01,3.338943862915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,8,1,power_law_1.2,1.7262079238891601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,8,power_law_1.2,0.1085055947303772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,8,power_law_1.2,0.2915776014328003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,8,power_law_1.2,0.10894720554351807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,8,power_law_1.2,0.11058559417724609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,8,power_law_1.2,0.30522239208221436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,8,power_law_1.2,0.11061760187149047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,8,power_law_1.2,0.114956796169281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,8,power_law_1.2,0.13463040590286254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,8,power_law_1.2,0.3566015958786011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,8,1,power_law_1.2,0.5550335884094239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,8,power_law_1.2,0.176147198677063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,8,power_law_1.2,0.18882559537887572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,8,power_law_1.2,0.36735999584198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,8,power_law_1.2,0.2408384084701538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,8,power_law_1.2,0.4182079792022705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,8,power_law_1.2,0.21990399360656737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,8,power_law_1.2,0.17244800329208373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,8,power_law_1.2,0.19029120206832886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,8,power_law_1.2,0.4512959957122803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,8,power_law_1.2,0.19864959716796876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,8,power_law_1.2,0.22651519775390624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,8,power_law_1.2,0.5486976146697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,8,power_law_1.2,0.25114879608154295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,8,power_law_1.2,0.2715775966644287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,8,power_law_1.01,0.323635196685791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,8,power_law_1.2,0.28752639293670657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,8,power_law_1.2,0.3482111930847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,8,power_law_1.2,0.45746560096740724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,8,power_law_1.2,0.7727615833282471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,8,power_law_1.2,1.0134079933166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,8,power_law_1.01,0.41429758071899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,8,power_law_1.2,0.6319424152374268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,8,power_law_1.01,9.030438232421876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,8,power_law_1.01,0.6057472229003906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,8,power_law_1.01,9.068556976318359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,8,power_law_1.2,1.3360063552856445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,8,power_law_1.01,4.875372695922851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,8,power_law_1.2,0.8728128433227539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,8,power_law_1.2,2.034886360168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,8,power_law_1.01,0.8191424369812011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,8,power_law_1.2,1.263814353942871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,8,power_law_1.2,2.5456192016601564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,8,power_law_1.01,13.274989318847656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,8,power_law_1.01,1.2526335716247559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,8,power_law_1.2,1.5255680084228516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,8,power_law_1.2,4.226950454711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,8,power_law_1.01,7.38656005859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,8,1,power_law_1.2,2.858425521850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,8,power_law_1.01,1.659641647338867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,8,power_law_1.2,2.522342491149902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,8,power_law_1.2,5.885260772705078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,8,power_law_1.2,0.05554559826850891
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,8,power_law_1.2,0.05566080212593079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,8,power_law_1.2,3.3544193267822267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,8,power_law_1.2,0.055731201171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,8,power_law_1.01,2.420595169067383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,8,power_law_1.2,0.055174398422241214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,8,power_law_1.01,18.805056762695312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,8,power_law_1.2,0.055929601192474365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,8,1,balanced,0.04791999856630961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,8,1,balanced,0.07005866865317027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,8,power_law_1.2,0.05937280058860779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,8,1,balanced,0.13062399625778198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,8,1,balanced,0.1264639993508657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,8,1,balanced,0.12923199931780496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,8,power_law_1.2,0.06728320121765137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,8,1,balanced,0.13149866461753845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,8,1,balanced,0.12745066483815512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,8,1,balanced,0.1260479986667633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,8,power_law_1.2,0.06679040193557739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,8,1,balanced,0.1386560002962748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,8,1,balanced,0.14467199643452963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,8,1,balanced,0.1346506675084432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,8,power_law_1.2,0.134662401676178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,8,1,balanced,0.13566399614016214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,8,1,balanced,0.13565867145856222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,8,1,balanced,0.14148799578348795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,8,power_law_1.2,0.10396159887313842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,8,1,balanced,0.20392000675201416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,8,1,balanced,0.2035520076751709
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,8,power_law_1.2,0.07461760044097901
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,8,power_law_1.2,0.14336639642715454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,8,power_law_1.2,0.1460543990135193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,8,power_law_1.2,0.09387519955635071
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,8,power_law_1.2,0.1450943946838379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,8,1,balanced,0.19817066192626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,8,1,balanced,0.22932799657185873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,8,1,balanced,0.23755200703938803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,8,power_law_1.2,0.1087615966796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,8,1,balanced,0.32546132802963257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,8,1,balanced,0.4149706761042277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,8,1,balanced,0.5947146813074747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,8,1,balanced,0.7370399634043375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,8,1,balanced,1.105679988861084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,8,power_law_1.2,0.1349120020866394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,8,1,balanced,1.393290678660075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,8,1,balanced,2.162090619405111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,8,power_law_1.2,0.1776128053665161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,8,1,balanced,2.8453760147094727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,8,1,balanced,0.04154666761557261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,8,1,balanced,0.06445866823196411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,8,1,balanced,0.08654933174451192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,8,1,balanced,0.08627733588218689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,8,1,balanced,0.08822932839393616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,8,1,balanced,0.09800533453623454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,8,1,balanced,0.11674666404724121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,8,1,balanced,0.12719466288884482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,8,1,balanced,0.12131200234095256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,8,1,balanced,0.14037332932154337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,8,1,balanced,0.09088533123334248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,8,1,balanced,0.09193600217501323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,8,1,balanced,0.09681066870689392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,8,1,balanced,0.0990826686223348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,8,1,balanced,0.13423466682434082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,8,1,balanced,0.1400266687075297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,8,1,balanced,0.13127467036247253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,8,1,balanced,0.15692800283432007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,8,1,balanced,0.16685867309570312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,8,1,balanced,0.22412800788879395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,8,1,balanced,0.2732906738917033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,8,1,balanced,0.3823466698328654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,8,1,balanced,0.470085342725118
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,8,power_law_1.2,0.2164031982421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,8,1,balanced,0.6842186450958252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,8,1,balanced,0.8803199927012125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,8,1,balanced,1.2981546719868977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,8,1,balanced,1.7017866770426433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,8,1,balanced,0.06394133468468984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,8,1,balanced,0.07392000158627827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,8,1,balanced,0.09157333771387736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,8,1,balanced,0.09059199690818787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,8,1,balanced,0.08950933814048767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,8,1,balanced,0.09497066338857015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,8,1,balanced,0.09713600079218547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,8,1,power_law_1.2,0.7313407897949219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,8,1,balanced,0.097461332877477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,8,1,balanced,0.1337386667728424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,8,1,balanced,0.1356160044670105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,8,1,balanced,0.13636266191800436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,8,1,balanced,0.13990400234858194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,8,1,balanced,0.1425279974937439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,8,1,balanced,0.14722133676211038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,8,1,balanced,0.21945067246754965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,8,1,balanced,0.22340265909830728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,8,1,balanced,0.22177600860595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,8,1,balanced,0.30196799834569293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,8,1,balanced,0.391210675239563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,8,1,balanced,0.5530080000559489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,8,1,balanced,0.6781013011932373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,8,1,balanced,1.0069119930267334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,8,1,balanced,1.2875839869181316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,8,1,balanced,1.8940693537394206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,8,1,balanced,2.442805290222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,8,1,balanced,3.600528081258138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,8,power_law_1.2,0.3018431901931763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,8,1,balanced,0.027450665831565857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,8,1,balanced,0.03342399994532267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,8,1,balanced,0.05373333394527435
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,8,1,balanced,0.0537013312180837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,8,1,balanced,0.052943999568621315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,8,1,balanced,0.05416533350944519
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,8,1,balanced,0.05917333563168844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,8,1,balanced,0.05990933378537496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,8,1,balanced,0.06379200021425883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,8,1,balanced,0.06443733473618825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,8,1,balanced,4.68501345316569
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,8,1,balanced,0.06307200094064076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,8,1,balanced,0.07032533486684163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,8,1,power_law_1.01,0.0466048002243042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,8,1,balanced,0.07085333267847697
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,8,1,balanced,0.07356800138950348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,8,1,power_law_1.01,0.06607999801635742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,8,1,balanced,0.09873066345850627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,8,1,balanced,0.08620799581209819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,8,1,power_law_1.01,0.0741375982761383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,8,1,balanced,0.08321600159009297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,8,1,balanced,0.14494933684666952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,8,1,power_law_1.01,0.12333439588546753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,8,1,balanced,0.11050132910410564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,8,1,power_law_1.01,0.12980480194091798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,8,1,power_law_1.01,0.13062399625778198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,8,power_law_1.01,12.871501159667968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,8,1,power_law_1.01,0.048716801404953006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,8,1,power_law_1.01,0.06343680024147033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,8,1,power_law_1.01,0.07018240094184876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,8,1,power_law_1.01,0.07642880082130432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,8,1,power_law_1.01,0.07646080255508422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,8,1,power_law_1.01,0.08261119723320007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,8,1,power_law_1.01,0.10085760354995728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,8,1,power_law_1.01,0.10708479881286621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,8,power_law_1.2,0.4277632236480713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,8,1,balanced,0.13838932911554971
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,8,1,balanced,0.1699840029080709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,8,1,power_law_1.01,0.11331199407577515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,8,1,balanced,0.23750933011372885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,8,1,power_law_1.01,0.1154047966003418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,8,1,power_law_1.01,0.12446080446243286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,8,1,balanced,0.30196799834569293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,8,1,power_law_1.01,0.09349120259284974
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,8,1,balanced,0.43805332978566486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,8,1,power_law_1.01,0.12446080446243286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,8,1,power_law_1.01,0.0974016010761261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,8,1,power_law_1.01,0.14140160083770753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,8,1,power_law_1.01,0.10059520006179809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,8,1,power_law_1.01,0.1417088031768799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,8,1,power_law_1.01,0.11469440460205078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,8,1,power_law_1.01,0.13763840198516847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,8,1,power_law_1.01,0.14841599464416505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,8,1,power_law_1.01,0.13019520044326782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,8,1,power_law_1.01,0.15040639638900757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,8,1,power_law_1.01,0.15525120496749878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,8,1,power_law_1.01,0.13337600231170654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,8,1,power_law_1.01,0.1918272018432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,8,1,power_law_1.01,0.13697919845581055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,8,1,balanced,0.569103995958964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,8,1,power_law_1.01,0.15134079456329347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,8,1,balanced,0.8270346323649088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,8,1,balanced,1.082133372624715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,8,1,power_law_1.01,0.060755199193954466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,8,1,power_law_1.01,0.06656000018119812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,8,1,power_law_1.01,0.07371519804000855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,8,1,power_law_1.01,0.08595839738845826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,8,1,power_law_1.01,0.19271039962768555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,8,1,power_law_1.01,0.08706560134887695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,8,1,power_law_1.01,0.20798718929290771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,8,1,power_law_1.01,0.09132800102233887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,8,1,power_law_1.01,0.09292160272598267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,8,1,power_law_1.01,0.10518399477005005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,8,1,power_law_1.01,0.21303040981292726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,8,1,power_law_1.01,0.1302016019821167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,8,1,power_law_1.01,0.23645439147949218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,8,1,power_law_1.01,0.14009599685668944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,8,power_law_1.2,0.5905663967132568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,8,1,power_law_1.01,0.1327232003211975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,8,1,power_law_1.01,0.18394240140914916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,8,1,power_law_1.01,0.18677120208740233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,8,1,power_law_1.01,0.2060352087020874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,8,1,power_law_1.01,0.2936448097229004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,8,1,power_law_1.01,0.21626880168914794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,8,1,power_law_1.01,0.2234816074371338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,8,1,power_law_1.01,0.24356479644775392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,8,1,power_law_1.01,0.2803071975708008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,8,power_law_1.2,5.084064102172851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,8,1,power_law_1.01,0.35765120983123777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,8,1,power_law_1.01,0.37512960433959963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,8,1,power_law_1.01,0.39645440578460694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,8,1,power_law_1.01,0.45379838943481443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,8,1,power_law_1.01,0.27905280590057374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,8,power_law_1.2,0.8260671615600585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,8,1,power_law_1.01,0.561407995223999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,8,1,power_law_1.01,0.6171008110046386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,8,power_law_1.2,0.15728000402450562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,8,power_law_1.2,0.15797760486602783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,8,1,power_law_1.01,0.39073278903961184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,8,power_law_1.2,0.15709439516067505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,8,power_law_1.2,0.15816320180892945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,8,power_law_1.2,0.1600000023841858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,8,power_law_1.2,0.1793727993965149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,8,power_law_1.2,0.2292095899581909
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,8,power_law_1.2,0.24557440280914306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,8,power_law_1.2,0.2723968029022217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,8,power_law_1.2,0.28346240520477295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,8,1,power_law_1.01,0.7948351860046386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,8,1,power_law_1.01,0.4970816135406494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,8,power_law_1.2,0.31411840915679934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,8,power_law_1.2,0.37173759937286377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,8,power_law_1.2,0.36068480014801024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,8,power_law_1.2,0.44292478561401366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,8,power_law_1.2,1.2021183967590332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,8,power_law_1.2,0.490937614440918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,8,power_law_1.2,0.5338560104370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,8,1,power_law_1.01,0.7001471996307373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,8,power_law_1.2,9.66302719116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,8,power_law_1.2,0.7644480228424072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,8,power_law_1.2,0.9724864006042481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,8,1,power_law_1.01,0.7250112056732178
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,8,power_law_1.01,3.395532989501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,8,power_law_1.2,1.3154175758361817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,8,1,power_law_1.01,1.0069184303283691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,8,1,power_law_1.01,0.02924799919128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,8,1,power_law_1.01,0.03389439880847931
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,8,1,power_law_1.01,0.0392767995595932
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,8,1,power_law_1.01,0.051123201847076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,8,1,power_law_1.01,0.05167999863624573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,8,1,power_law_1.01,0.05258880257606506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,8,1,power_law_1.01,0.0579584002494812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,8,power_law_1.2,1.8727935791015624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,8,1,power_law_1.01,0.06423680186271667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,8,1,power_law_1.01,0.06032639741897583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,8,1,power_law_1.01,0.061452800035476686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,8,1,power_law_1.01,0.06794880032539367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,8,1,power_law_1.01,0.06714879870414733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,8,1,power_law_1.01,0.06778240203857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,8,1,power_law_1.01,0.07905279994010925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,8,1,power_law_1.01,0.07715200185775757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,8,1,power_law_1.01,0.08160640001296997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,8,1,power_law_1.01,0.09006080031394958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,8,1,power_law_1.01,0.1080896019935608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,8,1,power_law_1.01,1.2888704299926759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,8,power_law_1.2,2.4829631805419923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,8,1,power_law_1.01,0.12554880380630493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,8,1,power_law_1.01,1.1370559692382813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,8,1,power_law_1.01,0.1546175956726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,8,1,power_law_1.01,0.9076607704162598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,8,power_law_1.2,7.2075645446777346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,8,1,power_law_1.2,0.04668160080909729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,8,1,power_law_1.2,0.06293119788169861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,8,1,power_law_1.2,0.07996799945831298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,8,1,power_law_1.2,0.12806400060653686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,8,1,power_law_1.2,0.12776960134506227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,8,1,power_law_1.01,0.17588479518890382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,8,1,power_law_1.2,0.13479679822921753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,8,1,power_law_1.2,0.1309183955192566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,8,1,power_law_1.2,0.12862720489501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,8,1,power_law_1.2,0.14302719831466676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,8,1,power_law_1.2,0.1415552020072937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,8,1,power_law_1.2,0.13570560216903688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,8,1,power_law_1.2,0.14839040040969848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,8,1,power_law_1.2,0.149619197845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,8,1,power_law_1.2,0.1548735976219177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,8,1,power_law_1.2,0.19956480264663695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,8,1,power_law_1.01,1.4908415794372558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,8,1,power_law_1.2,0.20517759323120116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,8,1,power_law_1.2,0.21052799224853516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,8,1,power_law_1.2,0.23264000415802003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,8,1,power_law_1.2,0.2976896047592163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,8,1,power_law_1.2,0.37102720737457273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,8,power_law_1.2,3.6777854919433595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,8,1,power_law_1.01,1.8664255142211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,8,power_law_1.2,1.7130048751831055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,8,1,power_law_1.01,0.2542464017868042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,8,1,power_law_1.01,1.3291647911071778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,8,1,power_law_1.01,2.421945571899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,8,1,power_law_1.2,0.43439359664916993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,8,1,power_law_1.01,0.30933759212493894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,8,1,power_law_1.01,2.163039970397949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,8,power_law_1.2,4.812326431274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,8,1,power_law_1.01,0.44118399620056153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,8,1,power_law_1.2,0.613644790649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,8,1,power_law_1.01,3.559417724609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,8,power_law_1.2,2.452544021606445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,8,1,power_law_1.2,0.7949632167816162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,8,power_law_1.2,6.956025695800781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,8,1,power_law_1.2,1.1203840255737305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,8,power_law_1.2,12.718873596191406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,8,1,power_law_1.2,0.046700799465179445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,8,1,power_law_1.2,0.059411197900772095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,8,1,power_law_1.2,0.06598399877548218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,8,1,power_law_1.2,0.0762112021446228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,8,1,power_law_1.2,0.07427840232849121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,8,1,power_law_1.2,0.08104959726333619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,8,1,power_law_1.2,0.10367360115051269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,8,1,power_law_1.2,0.10526080131530761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,8,1,power_law_1.2,0.11469440460205078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,8,1,power_law_1.01,0.581983995437622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,8,1,power_law_1.2,0.1161728024482727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,8,1,power_law_1.01,0.8348544120788575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,8,1,power_law_1.2,0.09281920194625855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,8,1,power_law_1.2,0.09837440252304078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,8,1,power_law_1.2,0.09949439764022827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,8,1,power_law_1.2,0.10286719799041748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,8,1,power_law_1.2,0.12373119592666626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,8,1,power_law_1.2,0.1327936053276062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,8,1,power_law_1.2,0.13925119638442993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,8,1,power_law_1.2,0.15608320236206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,8,1,power_law_1.01,4.702719879150391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,8,1,power_law_1.2,0.06021760106086731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,8,1,power_law_1.01,2.8452096939086915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,8,1,power_law_1.2,0.06758400201797485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,8,1,power_law_1.2,0.07011839747428894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,8,1,power_law_1.2,0.08514559864997864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,8,1,power_law_1.2,0.08562560081481933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,8,1,power_law_1.2,0.09100160002708435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,8,1,power_law_1.2,0.09244160056114196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,8,1,power_law_1.2,0.10882560014724732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,8,1,power_law_1.2,0.19545600414276124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,8,1,power_law_1.2,0.12947839498519897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,8,1,power_law_1.2,0.1402176022529602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,8,1,power_law_1.01,1.108518409729004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,8,1,power_law_1.2,0.14361599683761597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,8,balanced,0.051674668987592064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,8,balanced,0.06582933167616527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,8,balanced,0.08964266379674275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,8,1,power_law_1.2,0.1843135952949524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,8,balanced,0.15828800201416016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,8,balanced,0.257152001063029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,8,balanced,0.4527999957402547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,8,1,power_law_1.2,0.18668160438537598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,8,balanced,0.48665066560109455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,8,balanced,0.48760533332824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,8,1,power_law_1.2,0.2037951946258545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,8,balanced,0.5060906807581583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,8,balanced,0.5105813344319662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,8,1,power_law_1.2,0.21615359783172608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,8,balanced,0.5678026676177979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,8,balanced,0.5723466475804647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,8,1,power_law_1.2,0.2236799955368042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,8,balanced,0.5783946514129639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,8,balanced,0.5893173217773438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,8,balanced,0.599397341410319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,8,1,power_law_1.2,0.279417610168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,8,balanced,0.6105493307113647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,8,balanced,0.6290293137232462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,8,balanced,0.7152799765268961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,8,balanced,0.7455519835154215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,8,1,power_law_1.2,0.0297791987657547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,8,1,power_law_1.2,0.03289600014686585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,8,balanced,0.8104266325632731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,8,1,power_law_1.2,0.3494976043701172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,8,1,power_law_1.2,0.036748799681663516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,8,1,power_law_1.2,0.2486720085144043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,8,balanced,0.8753226598103842
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,8,1,power_law_1.2,0.05292159914970398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,8,1,power_law_1.2,0.05297279953956604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,8,balanced,1.0035999615987141
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,8,1,power_law_1.2,0.055955201387405396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,8,balanced,1.1365066369374592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,8,1,power_law_1.2,0.05867519974708557
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,8,1,power_law_1.2,0.06403200030326843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,8,balanced,1.67194668451945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,8,1,power_law_1.2,0.06005120277404785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,8,1,power_law_1.2,0.061926400661468504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,8,1,power_law_1.2,1.4724415779113769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,8,1,power_law_1.2,0.39812479019165037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,8,balanced,2.0443414052327475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,8,balanced,2.823989232381185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,8,balanced,3.641178766886393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,8,1,power_law_1.2,0.2854655981063843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,8,balanced,0.039850667119026184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,8,1,power_law_1.2,0.06776959896087646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,8,balanced,0.048165331284205117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,8,balanced,0.06966400146484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,8,balanced,0.09542933106422424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,8,1,power_law_1.2,0.0680512011051178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,8,balanced,0.15440000096956888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,8,balanced,0.2733173370361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,8,1,power_law_1.2,0.07012479901313781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,8,1,power_law_1.2,0.5683519840240479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,8,balanced,0.30828799804051715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,8,1,power_law_1.01,1.730156707763672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,8,balanced,0.3139573335647583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,8,balanced,0.2892426649729411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,8,balanced,0.2942240039507548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,8,1,power_law_1.2,0.07872639894485474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,8,balanced,0.3577920198440552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,8,balanced,0.36906667550404865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,8,balanced,0.3764266570409139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,8,balanced,0.3856000105539958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,8,1,power_law_1.2,0.0777728021144867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,8,balanced,0.4004426797231038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,8,balanced,0.41238399346669513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,8,1,power_law_1.2,0.08414720296859741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,8,balanced,0.4341866572697957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,8,balanced,0.5004800160725912
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,8,1,power_law_1.2,0.09079679846763611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,8,balanced,0.5331360101699829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,8,balanced,0.6044960021972656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,8,balanced,0.6740640004475912
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,8,1,power_law_1.2,0.10280319452285766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,8,balanced,0.8114613691965739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,8,balanced,0.952186663945516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,8,1,power_law_1.2,0.7257343769073487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,8,balanced,1.3769013086954753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,8,balanced,1.6607732772827148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,8,balanced,2.372426668802897
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,8,1,power_law_1.2,0.12632319927215577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,8,1,power_law_1.2,0.39079039096832274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,8,balanced,3.067898750305176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,8,1,power_law_1.2,1.0114432334899903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,8,1,power_law_1.2,0.15086719989776612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,8,power_law_1.2,9.574848175048828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,8,balanced,0.030447999636332195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,8,balanced,0.031930667658646904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,8,balanced,0.040063999593257904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,8,balanced,0.062165334820747375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,8,balanced,0.09081066648165385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,8,balanced,0.14589333534240723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,8,balanced,0.15093333522478738
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,8,balanced,0.15457066893577576
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,8,balanced,0.15184533596038818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,8,balanced,0.15292267004648843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,8,balanced,0.060138667623202004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,8,1,power_law_1.2,0.18525439500808716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,8,balanced,0.15624533096949259
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,8,balanced,0.15524799625078836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,8,balanced,0.16080533464749655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,8,1,power_law_1.2,0.5037695884704589
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,8,balanced,0.16429866353670755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,8,balanced,0.21107200781504312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,8,balanced,0.20270933707555136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,8,balanced,0.18931732575098673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,8,power_law_1.2,3.3315006256103517
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,8,balanced,0.35738666852315265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,8,balanced,0.2789280017217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,8,balanced,0.22919466098149618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,8,power_law_1.01,0.17498879432678222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,8,balanced,0.24834666649500528
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,8,balanced,0.28947200377782184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,8,power_law_1.01,0.27631359100341796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,8,balanced,0.33429865042368573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,8,balanced,0.478005329767863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,8,power_law_1.01,0.31315200328826903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,8,balanced,0.5650399923324585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,8,power_law_1.01,0.32364161014556886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,8,balanced,0.8087360064188639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,8,1,power_law_1.2,1.2950271606445312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,8,power_law_1.01,0.37421441078186035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,8,balanced,1.0567039648691814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,8,power_law_1.01,0.40621438026428225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,8,power_law_1.01,0.08831999897956848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,8,power_law_1.01,0.38894081115722656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,8,power_law_1.01,0.14914560317993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,8,power_law_1.01,0.16955519914627076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,8,power_law_1.01,0.401964807510376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,8,power_law_1.01,0.21994879245758056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,8,power_law_1.01,0.46535677909851075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,8,power_law_1.01,0.24043519496917726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,8,power_law_1.01,0.45674881935119627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,8,power_law_1.01,0.26184959411621095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,8,power_law_1.01,0.4665215969085693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,8,power_law_1.01,0.26832640171051025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,8,power_law_1.01,0.5380928039550781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,8,power_law_1.01,0.28844799995422366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,8,power_law_1.01,0.5426623821258545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,8,power_law_1.01,0.2794303894042969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,8,power_law_1.01,0.5337215900421143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,8,power_law_1.01,0.2768512010574341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,8,power_law_1.01,0.5590847969055176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,8,power_law_1.01,0.33609600067138673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,8,power_law_1.01,0.5674880027770997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,8,power_law_1.01,0.33710079193115233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,8,power_law_1.01,0.3413568019866943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,8,power_law_1.01,0.5895487785339355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,8,power_law_1.01,0.35825281143188475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,8,1,power_law_1.2,2.156287956237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,8,power_law_1.01,0.7094207763671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,8,power_law_1.01,0.3878592014312744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,8,power_law_1.01,0.7427135944366455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,8,power_law_1.01,0.39013121128082273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,8,balanced,0.06842133402824402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,8,power_law_1.01,0.8514176368713379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,8,1,power_law_1.2,0.25340800285339354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,8,power_law_1.01,0.9887104034423828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,8,power_law_1.01,1.1999808311462403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,8,power_law_1.01,1.4742464065551757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,8,power_law_1.01,0.419814395904541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,8,power_law_1.01,0.4853248119354248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,8,power_law_1.01,1.8914623260498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,8,power_law_1.01,0.5374335765838623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,8,power_law_1.01,0.6278143882751465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,8,power_law_1.01,0.7285312175750732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,8,power_law_1.01,2.4466495513916016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,8,power_law_1.01,0.9313792228698731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,8,power_law_1.01,1.107487964630127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,8,power_law_1.01,3.485977554321289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,8,power_law_1.01,1.5212096214294433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,8,power_law_1.01,1.9028480529785157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,8,power_law_1.01,4.588940811157227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,8,power_law_1.01,2.77587833404541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,8,1,power_law_1.2,1.866694450378418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,8,1,power_law_1.2,0.6990848064422608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,8,power_law_1.01,3.551839828491211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,8,1,power_law_1.2,0.3166271924972534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,8,balanced,0.08257600168387096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,8,power_law_1.01,0.05711359977722168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,8,power_law_1.01,0.08409600257873535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,8,power_law_1.01,0.10096640586853027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,8,power_law_1.01,0.10116480588912964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,8,power_law_1.01,0.12272000312805176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,8,power_law_1.01,0.12128000259399414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,8,power_law_1.01,0.14337919950485228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,8,power_law_1.01,0.14133119583129883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,8,power_law_1.01,0.13864959478378297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,8,power_law_1.01,0.139136004447937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,8,power_law_1.01,0.14639999866485595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,8,power_law_1.01,0.14428800344467163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,8,power_law_1.01,0.1475775957107544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,8,power_law_1.01,0.1509119987487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,8,1,power_law_1.2,2.4227519989013673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,8,power_law_1.01,0.17144320011138917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,8,power_law_1.01,0.17165440320968628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,8,power_law_1.01,0.0967423975467682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,8,power_law_1.01,0.17549439668655395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,8,power_law_1.01,0.2388672113418579
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,8,power_law_1.01,0.23952000141143798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,8,power_law_1.01,0.24111359119415282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,8,power_law_1.01,0.27489280700683594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,8,power_law_1.01,0.34079999923706056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,8,power_law_1.01,0.4188543796539307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,8,power_law_1.01,0.5639552116394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,8,balanced,0.11100799838701884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,8,balanced,0.1602133313814799
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,8,power_law_1.01,0.7140799999237061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,8,balanced,0.2802720069885254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,8,balanced,0.28173865874608356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,8,power_law_1.01,1.0153599739074708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,8,power_law_1.01,1.545785617828369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,8,power_law_1.2,0.15591039657592773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,8,power_law_1.2,0.2465536117553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,8,1,power_law_1.2,0.4424704074859619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,8,power_law_1.2,0.2414720058441162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,8,power_law_1.2,0.3017791986465454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,8,power_law_1.2,0.37077760696411133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,8,power_law_1.2,0.4013951778411865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,8,power_law_1.2,0.38540799617767335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,8,power_law_1.2,0.41125121116638186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,8,power_law_1.2,0.4454783916473389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,8,power_law_1.2,0.47084798812866213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,8,power_law_1.2,0.4754496097564697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,8,1,power_law_1.2,2.8220800399780273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,8,power_law_1.2,0.10033919811248779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,8,power_law_1.2,0.15812480449676514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,8,power_law_1.2,0.1458624005317688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,8,power_law_1.2,0.2128063917160034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,8,power_law_1.2,0.23297278881072997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,8,power_law_1.2,0.48862719535827637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,8,power_law_1.2,0.25886080265045164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,8,power_law_1.2,0.5234367847442627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,8,power_law_1.2,0.25338881015777587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,8,power_law_1.01,0.1601088047027588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,8,power_law_1.2,0.25813119411468505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,8,power_law_1.2,0.2669248104095459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,8,power_law_1.2,0.26747519969940187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,8,power_law_1.2,0.3304768085479736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,8,power_law_1.2,0.35418241024017333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,8,power_law_1.2,0.36071679592132566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,8,power_law_1.2,0.355020809173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,8,power_law_1.2,0.5521215915679931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,8,power_law_1.2,0.3836544036865234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,8,power_law_1.2,0.548857593536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,8,power_law_1.2,0.3974463939666748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,8,balanced,0.28600533803304035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,8,balanced,0.2905386686325073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,8,power_law_1.2,0.5844223976135254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,8,power_law_1.2,0.4220736026763916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,8,power_law_1.2,0.6032896041870117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,8,power_law_1.2,0.501145601272583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,8,power_law_1.2,0.5469696044921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,8,power_law_1.2,0.7165311813354492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,8,power_law_1.2,0.6679872035980224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,8,power_law_1.2,0.7788608074188232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,8,power_law_1.2,0.741107177734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,8,power_law_1.2,0.8929408073425293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,8,power_law_1.2,0.9563839912414551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,8,power_law_1.2,1.0405376434326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,8,power_law_1.2,1.177177619934082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,8,power_law_1.2,1.266214370727539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,8,power_law_1.2,1.6079935073852538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,8,power_law_1.2,1.556499195098877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,8,power_law_1.2,1.971104049682617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,8,power_law_1.2,2.0960128784179686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,8,power_law_1.2,2.9606719970703126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,8,power_law_1.2,2.6757696151733397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,8,power_law_1.2,3.708710479736328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,8,power_law_1.2,3.856403350830078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,8,power_law_1.2,5.010943984985351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,8,power_law_1.01,0.18971519470214843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,8,1,power_law_1.2,0.5758016109466553
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,8,power_law_1.2,0.0573311984539032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,8,power_law_1.2,0.08269439935684204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,8,power_law_1.2,0.08648959994316101
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,8,power_law_1.2,0.10325759649276733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,8,power_law_1.2,0.11351679563522339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,8,1,power_law_1.2,0.9099776268005371
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,8,power_law_1.2,0.13185280561447144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,8,power_law_1.2,0.13742719888687133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,8,power_law_1.2,0.1346560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,8,power_law_1.2,0.13715840578079225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,8,power_law_1.2,0.13306879997253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,8,power_law_1.2,0.13929599523544312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,8,balanced,0.5019893248875936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,8,power_law_1.2,0.09403520226478576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,8,balanced,0.5082613229751587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,8,balanced,0.5161173343658447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,8,balanced,0.5245493253072103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,8,power_law_1.2,0.1443711996078491
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,8,power_law_1.2,0.14218239784240722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,8,power_law_1.01,0.23529601097106934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,8,power_law_1.2,0.15697280168533326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,8,power_law_1.01,0.24027519226074218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,8,power_law_1.2,0.17089920043945311
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,8,power_law_1.2,0.1664255976676941
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,8,power_law_1.2,0.17777279615402222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,8,power_law_1.2,0.2648832082748413
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,8,power_law_1.2,0.23853440284729005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,8,power_law_1.2,0.2516479969024658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,8,power_law_1.2,0.2959167957305908
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,8,power_law_1.2,0.3671231985092163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,8,power_law_1.01,0.2721407890319824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,8,power_law_1.01,0.2668287992477417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,8,power_law_1.2,0.4494272232055664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,8,power_law_1.2,0.5958847999572754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,8,power_law_1.2,0.7738560199737549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,8,power_law_1.2,14.398265075683593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,8,power_law_1.2,1.0883584022521973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,8,power_law_1.2,1.7953407287597656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,8,1,balanced,0.04373333354791006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,8,1,balanced,0.05797333518664042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,8,1,balanced,0.08870399991671245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,8,1,balanced,0.17043733596801758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,8,1,balanced,0.3012160062789917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,8,1,balanced,0.550597349802653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,8,1,balanced,0.5566186507542928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,8,1,balanced,0.5578666528066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,8,power_law_1.2,0.1506880044937134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,8,1,balanced,0.5020373264948527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,8,1,balanced,0.5040160020192465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,8,1,balanced,0.5528746843338013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,8,1,balanced,0.5486773252487183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,8,1,balanced,0.5522720019022623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,8,1,balanced,0.5596693356831869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,8,1,balanced,0.5678079922993978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,8,1,balanced,0.5741013288497925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,8,1,balanced,0.5867733160654703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,8,1,balanced,0.7319626808166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,8,1,balanced,0.7507733503977457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,8,1,balanced,0.7904480298360189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,8,1,power_law_1.2,3.569478225708008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,8,1,balanced,0.8329973220825195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,8,1,balanced,0.9214133421579996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,8,1,balanced,1.0107253392537434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,8,1,balanced,1.531061331431071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,8,1,balanced,1.720080057779948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,8,1,balanced,2.4845546086629233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,8,1,balanced,3.2561705907185874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,8,1,balanced,0.03565866748491923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,8,1,balanced,0.04796266555786133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,8,1,balanced,0.06527466575304668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,8,power_law_1.01,0.24622080326080323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,8,1,balanced,0.09702400366465251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,8,1,balanced,0.1734880010286967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,8,1,balanced,0.3056959907213847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,8,power_law_1.01,0.2611583948135376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,8,1,balanced,0.36184000968933105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,8,1,balanced,0.3646186590194702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,8,1,balanced,0.3275253375371297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,8,1,balanced,0.34007465839385986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,8,1,balanced,0.3622560103734334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,8,1,balanced,0.3608106772104899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,8,1,balanced,0.36380799611409503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,8,1,balanced,0.37247467041015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,8,1,balanced,0.3785119851430257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,8,1,balanced,0.38471468289693195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,8,1,balanced,0.3970133463541667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,8,1,balanced,0.5204853216807047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,8,1,balanced,0.5405973196029663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,8,balanced,0.554640014966329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,8,1,balanced,0.582640012105306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,8,balanced,0.5675679842631022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,8,1,balanced,0.6308533350626627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,8,balanced,0.5805546840031942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,8,1,balanced,0.7189493179321289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,8,1,balanced,0.8102400302886963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,8,1,balanced,1.2233013312021892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,8,1,balanced,1.403557300567627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,8,1,balanced,2.0260693232218423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,8,power_law_1.2,0.1739776015281677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,8,1,balanced,2.611599922180176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,8,1,power_law_1.2,0.8395008087158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,8,1,balanced,0.04374399781227112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,8,power_law_1.2,0.20650238990783693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,8,power_law_1.2,0.2058880090713501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,8,power_law_1.2,0.21639680862426758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,8,power_law_1.2,0.2567552089691162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,8,power_law_1.01,0.45372800827026366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,8,power_law_1.01,0.4632256031036377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,8,power_law_1.01,0.4806272029876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,8,power_law_1.01,0.4904064178466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,8,balanced,0.6012906630833944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,8,balanced,0.6480426788330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,8,1,balanced,0.06026133398214976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,8,power_law_1.2,18.697369384765626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,8,power_law_1.2,0.25395200252532957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,8,power_law_1.2,0.2721856117248535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,8,1,balanced,0.031210665901501972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,8,1,balanced,0.033626665671666466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,8,1,balanced,0.042026668787002563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,8,1,balanced,0.06295999884605408
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,8,1,balanced,0.0942026674747467
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,8,1,balanced,0.15497066577275595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,8,1,balanced,0.15987733006477356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,8,1,balanced,0.163674662510554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,8,1,balanced,0.16590399543444315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,8,1,balanced,0.16940800348917642
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,8,1,balanced,0.17283199230829874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,8,1,balanced,0.18096532424290976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,8,1,balanced,0.18638400236765543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,8,1,balanced,0.19529066483179727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,8,1,balanced,0.22246400515238443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,8,1,balanced,0.23269333442052206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,8,1,balanced,0.250709335009257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,8,1,balanced,0.2582239905993144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,8,1,balanced,0.2822293241818746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,8,1,balanced,0.31409599383672077
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,8,1,balanced,0.3511199951171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,8,1,balanced,0.4458133379618327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,8,1,balanced,0.08685333530108134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,8,1,balanced,0.5257120132446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,8,1,balanced,0.7798933188120524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,8,1,balanced,0.7459306716918945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,8,1,balanced,1.061247984568278
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,8,1,balanced,1.384325345357259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,8,1,power_law_1.01,0.04240640103816986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,8,1,power_law_1.01,0.057120001316070555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,8,1,power_law_1.2,1.3180095672607421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,8,1,power_law_1.01,0.0862272024154663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,8,1,power_law_1.01,0.16733440160751342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,8,1,power_law_1.01,0.23162240982055665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,8,1,power_law_1.01,0.30992000102996825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,8,1,power_law_1.01,0.41884799003601075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,8,1,power_law_1.01,0.4257984161376953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,8,balanced,0.6955626805623373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,8,1,power_law_1.01,0.43413119316101073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,8,1,power_law_1.01,0.4396416187286377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,8,1,power_law_1.01,0.45361919403076173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,8,1,power_law_1.01,0.4621888160705566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,8,1,power_law_1.01,0.48593921661376954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,8,1,power_law_1.01,0.5167103767395019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,8,power_law_1.01,0.5287487983703614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,8,1,power_law_1.01,0.5125696182250976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,8,power_law_1.01,0.5368576049804688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,8,1,power_law_1.01,0.5283584117889404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,8,power_law_1.01,0.5485119819641113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,8,1,power_law_1.01,0.5622464179992676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,8,1,power_law_1.01,0.7096000194549561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,8,1,power_law_1.01,0.7341440200805665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,8,1,power_law_1.01,0.7945919990539551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,8,1,power_law_1.01,0.9195520401000976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,8,1,balanced,0.11029332876205444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,8,1,power_law_1.01,1.0865663528442382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,8,1,power_law_1.01,1.248953628540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,8,1,power_law_1.01,1.619366455078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,8,1,power_law_1.2,4.6831615447998045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,8,power_law_1.2,0.4663680076599121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,8,1,power_law_1.01,0.031814399361610415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,8,1,power_law_1.01,0.03976320028305054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,8,power_law_1.2,0.44136958122253417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,8,1,power_law_1.01,1.9786432266235352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,8,1,power_law_1.01,0.06110720038414001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,8,power_law_1.2,0.46518402099609374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,8,1,power_law_1.01,0.09071999788284302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,8,power_law_1.2,0.4760447978973389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,8,1,power_law_1.01,0.13804160356521605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,8,1,power_law_1.01,2.704595184326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,8,1,power_law_1.01,0.17155200242996216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,8,1,power_law_1.01,0.2675391912460327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,8,1,power_law_1.01,0.2784832000732422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,8,1,power_law_1.01,0.2849728107452393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,8,1,power_law_1.01,0.2867392063140869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,8,1,power_law_1.01,3.45560302734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,8,1,power_law_1.01,0.30108160972595216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,8,1,power_law_1.01,0.30529921054840087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,8,1,power_law_1.01,0.31128320693969724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,8,1,power_law_1.01,0.3285376071929932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,8,1,power_law_1.01,0.3481087923049927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,8,1,power_law_1.01,0.35966079235076903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,8,1,power_law_1.01,0.38200318813323975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,8,1,power_law_1.01,0.49419522285461426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,8,1,power_law_1.01,0.5393663883209229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,8,1,power_law_1.01,0.5976448059082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,8,1,power_law_1.01,0.6854080200195313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,8,1,power_law_1.01,0.8282303810119629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,8,1,power_law_1.01,0.9741951942443847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,8,1,power_law_1.01,1.2762751579284668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,8,1,balanced,0.21594132979710898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,8,1,power_law_1.01,1.5756416320800781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,8,1,power_law_1.01,2.1726207733154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,8,balanced,0.8185813426971436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,8,balanced,0.8941386540730795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,8,1,power_law_1.01,2.7756032943725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,8,1,power_law_1.01,0.04230400025844574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,8,power_law_1.01,0.6110720157623291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,8,power_law_1.01,0.68853759765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,8,1,power_law_1.01,0.02736639976501465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,8,1,power_law_1.01,0.029395198822021483
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,8,1,power_law_1.01,0.03993600010871887
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,8,1,power_law_1.01,0.059564799070358276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,8,1,power_law_1.01,0.07819520235061646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,8,1,power_law_1.01,0.10788480043411255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,8,1,power_law_1.01,0.13297280073165893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,8,1,power_law_1.01,0.13998080492019654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,8,1,power_law_1.01,0.14531199932098388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,8,1,power_law_1.01,0.14809600114822388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,8,1,power_law_1.01,0.15789439678192138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,8,power_law_1.2,0.5108928203582763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,8,1,power_law_1.01,0.1653439998626709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,8,power_law_1.2,0.5667456150054931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,8,power_law_1.2,0.576262378692627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,8,1,power_law_1.01,0.1744704008102417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,8,1,power_law_1.01,0.18899840116500854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,8,1,power_law_1.01,0.20817279815673828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,8,1,power_law_1.01,0.22513918876647948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,8,1,power_law_1.01,0.25911679267883303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,8,1,power_law_1.01,0.2526400089263916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,8,1,power_law_1.01,0.296729588508606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,8,1,balanced,0.3404639959335327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,8,1,power_law_1.01,0.058815997838973996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,8,1,balanced,0.34300267696380615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,8,1,power_law_1.01,0.31515519618988036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,8,1,power_law_1.01,0.37475199699401857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,8,1,power_law_1.01,0.4375487804412842
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,8,1,power_law_1.01,0.5357888221740723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,8,1,power_law_1.01,0.648518419265747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,8,1,power_law_1.01,0.8126655578613281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,8,1,power_law_1.01,1.1237631797790528
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,8,1,power_law_1.01,1.5330752372741698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,8,1,power_law_1.2,0.040556800365447995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,8,1,power_law_1.2,0.05699840188026428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,8,1,power_law_1.2,0.08677120208740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,8,1,power_law_1.2,0.15985920429229736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,8,1,power_law_1.2,0.2056704044342041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,8,balanced,1.3636800448099773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,8,1,power_law_1.2,0.2919552087783813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,8,balanced,1.558293342590332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,8,1,power_law_1.2,0.4136767864227295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,8,1,power_law_1.2,0.424505615234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,8,balanced,2.2386879920959473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,8,1,power_law_1.2,0.43096318244934084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,8,1,power_law_1.2,1.7228288650512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,8,1,power_law_1.2,0.43561601638793945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,8,1,power_law_1.2,0.03167999982833862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,8,balanced,2.913466771443685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,8,1,power_law_1.2,0.44965758323669436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,8,1,power_law_1.2,0.043110400438308716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,8,1,power_law_1.2,0.061343997716903687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,8,1,power_law_1.2,0.08766720294952393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,8,1,power_law_1.2,0.12761600017547609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,8,balanced,4.237194697062175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,8,1,power_law_1.2,0.16799360513687134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,8,1,power_law_1.2,0.2655807971954346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,8,power_law_1.01,0.7930880069732666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,8,1,power_law_1.2,0.2762239933013916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,8,1,power_law_1.2,0.4710847854614258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,8,1,power_law_1.2,0.288921594619751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,8,balanced,5.499935785929362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,8,1,power_law_1.2,0.49585919380187987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,8,1,power_law_1.2,0.2933120012283325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,8,1,power_law_1.2,0.2970815896987915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,8,1,power_law_1.2,0.5004672050476074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,8,1,power_law_1.2,0.3088831901550293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,8,1,power_law_1.2,0.5255040168762207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,8,1,power_law_1.01,0.08613119721412658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,8,1,power_law_1.2,0.31882240772247317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,8,1,power_law_1.2,0.5454463958740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,8,1,power_law_1.2,0.3515968084335327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,8,1,power_law_1.2,0.5644927978515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,8,1,power_law_1.2,0.36262400150299073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,8,1,power_law_1.2,0.7208255767822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,8,1,power_law_1.2,0.3698751926422119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,8,1,power_law_1.2,0.38549759387969973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,8,1,power_law_1.2,0.7581183910369873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,8,1,power_law_1.2,0.5187903881072998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,8,1,power_law_1.2,0.852729606628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,8,1,power_law_1.2,0.5477375984191895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,8,1,power_law_1.2,0.9392959594726562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,8,1,power_law_1.2,0.6339200019836426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,8,1,power_law_1.2,0.7023871898651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,8,1,power_law_1.2,1.1154623985290528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,8,power_law_1.2,0.6051775932312011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,8,1,power_law_1.2,0.855129623413086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,8,1,power_law_1.2,1.2979328155517578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,8,power_law_1.2,0.7571775913238525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,8,1,power_law_1.2,1.000755214691162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,8,1,power_law_1.2,1.6444736480712892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,8,1,power_law_1.2,1.2961088180541993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,8,1,power_law_1.2,2.0064064025878907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,8,1,power_law_1.2,1.596288013458252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,8,1,balanced,0.3444426854451497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,8,1,balanced,0.3439519802729289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,8,1,power_law_1.2,2.2040000915527345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,8,1,power_law_1.2,2.7503103256225585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,8,1,power_law_1.2,2.808870315551758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,8,1,power_law_1.2,3.4994945526123047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,8,balanced,0.04716266691684723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,8,balanced,0.06334400177001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,8,balanced,0.0786240001519521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,8,balanced,0.13993066549301147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,8,balanced,0.2243573268254598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,8,balanced,0.3948586781819661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,8,balanced,0.4233013391494751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,8,balanced,0.423253337542216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,8,balanced,0.4283039967219035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,8,1,power_law_1.2,0.04176000058650971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,8,balanced,0.43008001645406085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,8,1,power_law_1.2,1.1097472190856934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,8,balanced,0.48369598388671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,8,balanced,0.03944533318281174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,8,balanced,0.04565866788228353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,8,balanced,0.4882773160934448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,8,balanced,0.06629333396752675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,8,balanced,0.08598933617273967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,8,balanced,0.49266668160756427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,8,balanced,0.14381333192189535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,8,balanced,0.24292266368865967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,8,balanced,0.5025493303934733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,8,balanced,0.2731519937515259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,8,balanced,0.5112266540527344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,8,balanced,0.27644266684850055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,8,balanced,0.2630773385365804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,8,balanced,0.5202026764551798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,8,balanced,0.26269867022832233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,8,balanced,0.5413599809010824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,8,balanced,0.3166666626930237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,8,balanced,0.3264799912770589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,8,balanced,0.6254133383433024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,8,balanced,0.33294934034347534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,8,balanced,0.6533759832382202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,8,balanced,0.7113386789957682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,8,1,power_law_1.01,0.10928640365600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,8,balanced,0.7692853609720866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,8,balanced,0.884943962097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,8,balanced,1.0050453344980876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,8,balanced,1.480629285176595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,8,1,power_law_1.2,0.02725760042667389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,8,balanced,0.34324800968170166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,8,balanced,1.729157288869222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,8,balanced,0.35381333033243817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,8,1,power_law_1.2,0.028915199637413024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,8,balanced,0.36501868565877277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,8,balanced,0.38399465878804523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,8,1,power_law_1.2,0.04126079976558685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,8,balanced,0.4474666515986125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,8,balanced,2.4802239735921225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,8,balanced,0.4772426684697469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,8,balanced,0.5418453216552734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,8,balanced,0.6077066659927368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,8,balanced,0.7331146399180094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,8,1,balanced,0.33213333288828534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,8,balanced,3.260863939921061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,8,1,balanced,0.33473066488901776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,8,balanced,0.861845334370931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,8,1,balanced,0.3379146655400594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,8,1,balanced,0.3423786560694377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,8,balanced,1.2425866921742756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,8,1,balanced,0.35121599833170575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,8,1,balanced,0.35957332452138263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,8,1,balanced,0.3673706849416097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,8,balanced,1.496016025543213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,8,1,balanced,0.38307201862335205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,8,1,balanced,0.5103520154953003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,8,1,power_law_1.2,0.05633280277252197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,8,1,balanced,0.539408008257548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,8,balanced,2.139450709025065
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,8,1,power_law_1.2,0.07474560141563416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,8,1,balanced,0.7617759704589844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,8,1,power_law_1.2,0.10286719799041748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,8,1,balanced,0.9970400333404541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,8,power_law_1.01,1.020787239074707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,8,1,power_law_1.2,0.13249919414520264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,8,balanced,2.7663679122924805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,8,1,power_law_1.2,0.139628803730011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,8,power_law_1.01,1.2583552360534669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,8,1,power_law_1.2,0.14154239892959594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,8,1,power_law_1.2,0.14797439575195312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,8,1,power_law_1.2,0.1569983959197998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,8,1,power_law_1.2,0.16538239717483522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,8,1,power_law_1.2,0.17550079822540282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,8,1,power_law_1.2,0.19201279878616334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,8,1,power_law_1.2,0.2084671974182129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,8,1,power_law_1.2,0.22665600776672362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,8,1,power_law_1.2,0.05860480070114136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,8,1,power_law_1.2,0.2655616044998169
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,8,1,power_law_1.2,0.25899519920349123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,8,1,power_law_1.2,0.30339200496673585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,8,1,power_law_1.2,0.3210047960281372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,8,1,power_law_1.2,0.3816767930984497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,8,1,power_law_1.2,0.44736638069152834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,8,power_law_1.2,0.852729606628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,8,balanced,0.030293333033720653
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,8,1,power_law_1.2,0.5779583930969239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,8,balanced,0.03239466746648153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,8,balanced,0.03745066622893015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,8,balanced,0.05782400071620941
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,8,balanced,0.08241066833337148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,8,balanced,0.1300159990787506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,8,1,power_law_1.2,0.6574848175048829
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,8,balanced,0.13277332981427512
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,8,balanced,0.1378506620724996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,8,balanced,0.1357973317305247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,8,1,power_law_1.01,0.15191680192947388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,8,balanced,0.13570132851600647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,8,balanced,0.13938666383425394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,8,1,power_law_1.2,0.8229887962341309
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,8,balanced,0.13818666338920593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,8,balanced,0.13981333374977112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,8,balanced,0.14597333470980325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,8,balanced,0.19381332397460938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,8,balanced,0.18289599816004434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,8,1,power_law_1.2,1.1415295600891113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,8,1,power_law_1.2,1.567187213897705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,8,power_law_1.01,0.13144960403442382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,8,power_law_1.01,0.21797120571136475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,8,power_law_1.01,0.2623039960861206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,8,balanced,0.16880534092585245
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,8,balanced,0.3288266658782959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,8,power_law_1.01,0.2563584089279175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,8,balanced,0.22558399041493735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,8,balanced,0.2022613286972046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,8,balanced,0.2197386622428894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,8,power_law_1.01,0.319814395904541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,8,balanced,0.2534826596577962
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,8,balanced,0.2940160036087036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,8,power_law_1.01,0.2998143911361694
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,8,balanced,0.4184533357620239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,8,balanced,0.49666666984558105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,8,power_law_1.01,0.39089279174804686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,8,balanced,0.7095519701639811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,8,power_law_1.01,0.3760256052017212
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,8,balanced,0.9256532986958822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,8,power_law_1.01,0.380678391456604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,8,power_law_1.01,0.4111807823181152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,8,balanced,0.05533333122730255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,8,power_law_1.01,0.43091840744018556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,8,power_law_1.01,0.43637762069702146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,8,power_law_1.01,0.4692351818084717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,8,power_law_1.01,0.4624063968658447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,8,power_law_1.01,0.4857215881347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,8,power_law_1.01,0.5090367794036865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,8,power_law_1.01,0.5118847846984863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,8,1,power_law_1.2,0.08520960211753845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,8,power_law_1.01,0.6136896133422851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,8,power_law_1.01,0.6452479839324952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,8,power_law_1.01,0.758457612991333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,8,power_law_1.01,0.8778688430786132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,8,power_law_1.01,0.07924479842185975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,8,balanced,0.06037333110968272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,8,power_law_1.01,0.13540480136871338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,8,power_law_1.01,0.14852479696273804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,8,power_law_1.01,0.18595839738845826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,8,power_law_1.01,1.0958144187927246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,8,power_law_1.01,0.1684864044189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,8,power_law_1.01,1.281657600402832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,8,power_law_1.01,0.19164799451828002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,8,power_law_1.01,1.6155519485473633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,8,power_law_1.01,0.24873600006103516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,8,1,balanced,2.096560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,8,power_law_1.01,0.23697280883789062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,8,power_law_1.01,1.7799999237060546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,8,power_law_1.01,0.2503551959991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,8,power_law_1.01,2.046700859069824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,8,1,balanced,2.591317335764567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,8,power_law_1.01,0.2396224021911621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,8,power_law_1.01,2.142246437072754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,8,power_law_1.01,0.30734078884124755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,8,power_law_1.01,0.29694080352783203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,8,1,balanced,3.901263872782389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,8,power_law_1.01,2.908352088928223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,8,power_law_1.01,0.3163327932357788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,8,power_law_1.01,0.32191359996795654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,8,power_law_1.01,3.1457792282104493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,8,power_law_1.01,0.34213120937347413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,8,power_law_1.01,0.3563391923904419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,8,1,balanced,5.111802736918132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,8,power_law_1.01,3.4996543884277345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,8,power_law_1.01,0.3724800109863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,8,power_law_1.2,1.11015043258667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,8,power_law_1.01,0.44177918434143065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,8,balanced,0.07660266757011414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,8,power_law_1.01,0.4784832000732422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,8,power_law_1.01,4.118163299560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,8,power_law_1.2,1.304089641571045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,8,power_law_1.01,0.5649280071258544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,8,power_law_1.01,0.6441984176635742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,8,1,balanced,7.539818445841472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,8,power_law_1.01,5.342835235595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,8,power_law_1.01,0.8316864013671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,8,1,power_law_1.2,0.10536320209503174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,8,1,power_law_1.01,0.18934400081634523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,8,power_law_1.01,1.0188480377197267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,8,1,power_law_1.01,0.24225280284881592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,8,power_law_1.01,1.3556415557861328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,8,power_law_1.01,1.709516716003418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,8,power_law_1.01,6.790656280517578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,8,1,balanced,9.941792170206705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,8,power_law_1.2,0.1336832046508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,8,power_law_1.01,2.4280895233154296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,8,power_law_1.2,0.21453440189361572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,8,power_law_1.2,0.2081216096878052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,8,power_law_1.2,0.28752639293670657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,8,power_law_1.2,0.2909503936767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,8,power_law_1.01,3.1373952865600585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,8,power_law_1.2,0.287443208694458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,8,power_law_1.2,0.07998719811439514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,8,power_law_1.2,0.3837696075439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,8,power_law_1.2,0.383955192565918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,8,power_law_1.2,0.37046399116516116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,8,power_law_1.2,0.40660481452941893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,8,power_law_1.2,0.43853440284729006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,8,power_law_1.2,0.44722561836242675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,8,power_law_1.2,0.1373247981071472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,8,power_law_1.2,0.48156161308288575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,8,power_law_1.2,0.1363711953163147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,8,power_law_1.2,0.48555521965026854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,8,power_law_1.2,0.18071680068969725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,8,power_law_1.2,0.47842559814453123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,8,power_law_1.01,0.05252479910850525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,8,power_law_1.2,0.20207359790802001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,8,power_law_1.2,0.49114241600036623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,8,power_law_1.01,0.07684479951858521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,8,power_law_1.2,0.20281600952148438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,8,power_law_1.2,0.5320127964019775
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,8,power_law_1.01,0.0877568006515503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,8,power_law_1.2,0.2191231966018677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,8,power_law_1.2,0.631935977935791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,8,power_law_1.01,0.10516480207443238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,8,power_law_1.2,0.21833600997924804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,8,power_law_1.2,0.22522239685058593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,8,power_law_1.2,0.6860864162445068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,8,power_law_1.2,0.23887999057769777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,8,power_law_1.2,0.8167743682861328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,8,balanced,0.09590400258700053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,8,balanced,0.14627200365066528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,8,balanced,0.24365333716074625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,8,balanced,0.24630399545033774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,8,power_law_1.2,0.8888575553894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,8,power_law_1.2,1.139948844909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,8,power_law_1.01,0.08520320057868958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,8,power_law_1.01,0.10656640529632569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,8,1,power_law_1.2,0.13923840522766112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,8,power_law_1.2,1.4312640190124513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,8,power_law_1.01,0.11795200109481811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,8,power_law_1.2,0.2888767957687378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,8,power_law_1.01,0.12632960081100464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,8,power_law_1.2,0.32108159065246583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,8,power_law_1.01,0.12021759748458863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,8,power_law_1.2,1.834726333618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,8,power_law_1.2,0.3244607925415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,8,power_law_1.01,0.1161471962928772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,8,power_law_1.2,0.32990078926086425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,8,power_law_1.01,0.11902079582214356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,8,power_law_1.2,0.3416383981704712
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,8,power_law_1.01,0.13136639595031738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,8,power_law_1.2,0.3501440048217773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,8,power_law_1.01,0.12670719623565674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,8,power_law_1.2,0.36884479522705077
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,8,power_law_1.01,0.12703360319137574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,8,power_law_1.2,0.4447743892669678
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,8,power_law_1.01,0.13201279640197755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,8,power_law_1.01,0.15360000133514404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,8,power_law_1.2,0.49820799827575685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,8,power_law_1.01,0.1521407961845398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,8,power_law_1.2,0.5812160015106201
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,8,power_law_1.01,0.1583359956741333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,8,power_law_1.2,2.250937652587891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,8,power_law_1.2,0.6647744178771973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,8,power_law_1.01,0.2291584014892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,8,power_law_1.01,0.21262080669403077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,8,power_law_1.2,0.846399974822998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,8,power_law_1.01,0.2131903886795044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,8,power_law_1.01,0.24260480403900148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,8,power_law_1.2,1.0035776138305663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,8,power_law_1.2,3.368947219848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,8,power_law_1.01,0.2983871936798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,8,power_law_1.2,1.413491153717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,8,1,power_law_1.01,0.2461632013320923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,8,power_law_1.01,0.36734719276428224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,8,power_law_1.2,1.6653375625610352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,8,1,power_law_1.01,0.26026880741119385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,8,power_law_1.01,0.4960319995880127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,8,power_law_1.2,1.7490432739257813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,8,power_law_1.01,0.6222400188446044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,8,power_law_1.2,2.089779281616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,8,power_law_1.01,0.8630080223083496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,8,power_law_1.2,4.1767936706542965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,8,power_law_1.2,2.6107456207275392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,8,power_law_1.2,3.09102725982666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,8,power_law_1.01,1.3424832344055175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,8,power_law_1.2,3.3301246643066404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,8,1,balanced,0.03751999884843826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,8,1,balanced,0.052442664901415505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,8,1,balanced,0.07863999903202057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,8,1,balanced,0.1463466684023539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,8,1,balanced,0.2629973292350769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,8,power_law_1.2,4.052844619750976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,8,1,balanced,0.47704533735911053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,8,1,balanced,0.4817013343175252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,8,balanced,0.24838932355244955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,8,1,balanced,0.48208534717559814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,8,balanced,0.24999467531840006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,8,1,balanced,0.43084267775217694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,8,1,balanced,0.4325439929962158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,8,1,balanced,0.47865601380666095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,8,1,balanced,0.47486400604248047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,8,power_law_1.01,0.14433280229568482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,8,1,balanced,0.4774399995803833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,8,1,balanced,0.4828000068664551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,8,1,balanced,0.4862560033798218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,8,1,balanced,0.49399999777475995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,8,power_law_1.2,5.760550308227539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,8,1,balanced,0.5041173299153646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,8,1,balanced,0.6325600147247314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,8,1,balanced,0.6478453477223715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,8,1,balanced,0.6833439668019613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,8,1,balanced,0.7203359603881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,8,1,balanced,0.7976213296254476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,8,1,balanced,0.8765172958374023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,8,1,balanced,1.331221342086792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,8,power_law_1.2,0.052160000801086424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,8,power_law_1.2,0.07560960054397584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,8,1,balanced,1.5000534057617188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,8,power_law_1.2,7.569100952148437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,8,power_law_1.2,0.07799040079116822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,8,1,balanced,0.031082667410373688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,8,1,balanced,0.03965333352486292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,8,1,balanced,0.05712533493836721
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,8,power_law_1.2,0.08974720239639282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,8,1,balanced,2.186917304992676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,8,1,balanced,0.08260799944400787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,8,1,balanced,0.14863999684651694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,8,1,balanced,0.2642506758371989
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,8,power_law_1.2,0.10425599813461303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,8,1,balanced,0.31301865975062054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,8,1,balanced,0.3131840030352275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,8,1,balanced,0.29337066411972046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,8,power_law_1.2,0.12711679935455322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,8,1,balanced,0.2946666677792867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,8,1,power_law_1.01,0.27802879810333253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,8,1,balanced,0.31432000796000165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,8,1,balanced,0.31453333298365277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,8,power_law_1.2,0.11422079801559448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,8,1,power_law_1.01,0.28752000331878663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,8,1,balanced,0.3163466652234395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,8,1,power_law_1.2,0.17006080150604247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,8,1,balanced,0.3222986658414205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,8,1,balanced,0.32893866300582886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,8,power_law_1.2,0.12028800249099732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,8,1,balanced,0.3336319923400879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,8,1,power_law_1.2,0.23517439365386963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,8,1,balanced,0.3452586730321248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,8,power_law_1.2,0.1136512041091919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,8,1,balanced,0.4535839955012004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,8,1,balanced,0.46969600518544513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,8,power_law_1.2,0.12221440076828002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,8,1,balanced,0.5103413263956705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,8,power_law_1.2,0.12297600507736206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,8,1,balanced,0.5507680177688599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,8,1,balanced,0.6275573174158732
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,8,power_law_1.2,0.12764159440994263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,8,1,balanced,2.791103998819987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,8,1,balanced,0.7091306845347086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,8,power_law_1.2,0.12673280239105225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,8,1,balanced,1.0676053365071614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,8,power_law_1.2,0.13923840522766112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,8,1,power_law_1.01,0.30164480209350586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,8,power_law_1.2,0.15413119792938232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,8,1,balanced,1.2275359630584717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,8,1,power_law_1.01,0.31660799980163573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,8,power_law_1.2,0.14643839597702027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,8,1,power_law_1.01,0.33634560108184813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,8,1,balanced,1.7805387179056804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,8,power_law_1.2,0.16401920318603516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,8,1,power_law_1.01,0.3739007949829102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,8,power_law_1.2,0.08647680282592773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,8,power_law_1.2,0.249619197845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,8,1,power_law_1.01,0.4015359878540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,8,1,balanced,2.289951960245768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,8,power_law_1.2,0.20799999237060546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,8,1,power_law_1.01,0.4090432167053223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,8,power_law_1.2,0.21523840427398683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,8,1,power_law_1.01,0.5287680149078369
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,8,power_law_1.2,0.25772159099578856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,8,1,power_law_1.01,0.6232448101043702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,8,power_law_1.2,0.335699200630188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,8,1,power_law_1.01,0.867193603515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,8,power_law_1.2,0.3803711891174316
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,8,power_law_1.2,0.5475520133972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,8,1,power_law_1.01,1.0813247680664062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,8,power_law_1.2,0.7017024040222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,8,power_law_1.01,0.1794111967086792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,8,power_law_1.2,0.9219519615173339
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,8,power_law_1.2,1.5840319633483886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,8,1,power_law_1.01,0.03588480055332184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,8,1,balanced,0.02754133443037669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,8,1,balanced,0.030192000170548756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,8,1,balanced,0.03363733241955439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,8,1,power_law_1.01,0.050963199138641356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,8,1,balanced,0.05570666491985321
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,8,1,balanced,0.08432533343633015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,8,1,balanced,0.13606933752695718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,8,1,power_law_1.01,0.07624959945678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,8,1,power_law_1.01,0.14540799856185913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,8,power_law_1.2,0.145798397064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,8,1,power_law_1.01,0.19646719694137574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,8,1,power_law_1.01,0.26933119297027586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,8,1,power_law_1.01,0.36350719928741454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,8,1,power_law_1.01,0.36789119243621826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,8,1,balanced,0.13799466689427695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,8,1,power_law_1.01,0.3762495994567871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,8,1,power_law_1.01,0.3773184061050415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,8,1,power_law_1.01,0.3883455991744995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,8,1,balanced,0.03987200061480204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,8,1,power_law_1.01,0.3990015983581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,8,1,power_law_1.01,0.40805759429931643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,8,1,power_law_1.01,0.44481282234191893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,8,1,balanced,0.14383467038472494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,8,1,power_law_1.01,0.4415296077728271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,8,1,balanced,0.14574399590492249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,8,1,balanced,0.1476533313592275
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,8,1,balanced,0.15211199720700583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,8,1,power_law_1.01,0.4581759929656982
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,8,1,balanced,0.15646933515866598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,8,1,balanced,0.16081066926320395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,8,1,balanced,0.17090133825937906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,8,1,power_law_1.01,0.4747647762298584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,8,1,balanced,0.1935946742693583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,8,1,balanced,0.2011680006980896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,8,1,balanced,0.21831466754277548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,8,1,power_law_1.01,0.606496000289917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,8,1,power_law_1.01,0.6380735874176026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,8,1,power_law_1.01,0.6906367778778076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,8,balanced,0.43532268206278485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,8,power_law_1.2,0.13752319812774658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,8,balanced,0.44281601905822754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,8,balanced,0.4492959976196289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,8,1,power_law_1.01,0.7929344177246094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,8,balanced,0.45681599775950116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,8,1,power_law_1.01,0.9494976043701172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,8,1,balanced,0.22338134050369263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,8,1,balanced,0.2449386715888977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,8,1,power_law_1.2,0.24563200473785402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,8,1,balanced,0.2718240022659302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,8,1,power_law_1.01,1.0796287536621094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,8,1,balanced,0.3053599993387858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,8,1,power_law_1.2,0.2539776086807251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,8,1,balanced,0.3926933209101359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,8,1,balanced,0.43281598885854083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,8,1,balanced,0.6995680332183838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,8,1,power_law_1.01,1.4045696258544922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,8,1,balanced,0.6509226560592651
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,8,1,balanced,0.9159946441650391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,8,1,power_law_1.01,1.7204479217529296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,8,1,balanced,1.192309300104777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,8,power_law_1.01,0.1836351990699768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,8,1,power_law_1.01,0.02959359884262085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,8,power_law_1.01,0.2028736114501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,8,1,power_law_1.01,0.038150399923324585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,8,1,power_law_1.01,2.343539237976074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,8,power_law_1.01,0.16933120489120485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,8,1,power_law_1.01,0.0543936014175415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,8,1,power_law_1.01,0.08039039969444275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,8,power_law_1.01,0.23508479595184326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,8,1,power_law_1.01,0.11781760454177856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,8,1,power_law_1.01,0.15775359869003297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,8,1,power_law_1.01,2.9856576919555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,8,1,power_law_1.01,0.2355072021484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,8,1,power_law_1.01,0.2427839994430542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,8,1,power_law_1.01,0.2477247953414917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,8,1,power_law_1.01,0.25474560260772705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,8,1,power_law_1.01,0.2698496103286743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,8,1,power_law_1.01,0.2682368040084839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,8,1,power_law_1.01,0.2775808095932007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,8,1,balanced,0.05611733098824819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,8,1,power_law_1.01,0.29367680549621583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,8,1,power_law_1.01,0.30305280685424807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,8,1,power_law_1.01,0.31007359027862547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,8,1,power_law_1.01,0.3316927909851074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,8,1,power_law_1.01,0.4376959800720215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,8,1,power_law_1.01,0.46681599617004393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,8,1,power_law_1.01,0.5203839778900147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,8,1,power_law_1.01,0.5999680042266846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,8,1,power_law_1.01,0.7181568145751953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,8,1,power_law_1.01,0.8402496337890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,8,1,power_law_1.01,1.1129983901977538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,8,power_law_1.2,0.18020479679107665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,8,power_law_1.2,0.21006081104278565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,8,1,power_law_1.01,1.3703167915344239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,8,power_law_1.2,0.2089855909347534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,8,power_law_1.2,0.23559041023254396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,8,1,power_law_1.2,0.2791872024536133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,8,1,power_law_1.01,1.898636817932129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,8,1,power_law_1.2,0.28936960697174074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,8,1,power_law_1.01,2.3648895263671874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,8,1,power_law_1.2,0.30512640476226804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,8,1,power_law_1.2,0.3153215885162354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,8,1,power_law_1.01,2.4323776245117186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,8,1,power_law_1.2,0.33489279747009276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,8,1,power_law_1.2,0.385913610458374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,8,1,power_law_1.01,2.9493440628051757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,8,1,power_law_1.2,0.4106880187988281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,8,1,power_law_1.2,0.4178112030029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,8,1,power_law_1.2,0.534329605102539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,8,power_law_1.01,0.23852159976959228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,8,1,power_law_1.01,4.145222473144531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,8,1,power_law_1.2,0.6587456226348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,8,power_law_1.01,0.2324608087539673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,8,1,power_law_1.2,0.8733056068420411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,8,1,power_law_1.2,1.1058943748474122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,8,1,power_law_1.01,0.03823359906673431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,8,1,power_law_1.01,5.299481582641602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,8,1,balanced,0.07842133442560832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,8,1,power_law_1.01,0.025721600651741026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,8,1,power_law_1.01,0.027846398949623107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,8,balanced,0.48125867048899335
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,8,1,power_law_1.01,0.03607040047645569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,8,balanced,0.49398934841156006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,8,balanced,0.5053386688232422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,8,1,power_law_1.01,0.05493760108947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,8,1,power_law_1.01,0.07002879977226258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,8,1,power_law_1.01,0.09667840003967285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,8,1,power_law_1.01,7.690739440917969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,8,1,power_law_1.01,0.11512960195541382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,8,1,power_law_1.01,0.12283519506454468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,8,1,power_law_1.01,0.1269503951072693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,8,1,power_law_1.01,0.13073920011520385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,8,1,power_law_1.01,0.13791999816894532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,8,1,power_law_1.01,0.14351999759674072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,8,1,power_law_1.01,0.1518336057662964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,8,1,power_law_1.01,0.16584320068359376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,8,1,power_law_1.01,0.1827008008956909
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,8,1,power_law_1.01,0.1946239948272705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,8,power_law_1.2,0.20087039470672607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,8,1,power_law_1.01,0.22704000473022462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,8,power_law_1.2,0.2385792016983032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,8,1,power_law_1.01,10.061260986328126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,8,1,power_law_1.01,0.2185920000076294
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,8,1,power_law_1.01,0.2580928087234497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,8,1,power_law_1.2,0.0358271986246109
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,8,1,power_law_1.01,0.2721343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,8,1,power_law_1.2,0.051737600564956666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,8,1,power_law_1.2,0.07703040242195129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,8,1,power_law_1.01,0.3255872011184692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,8,1,power_law_1.2,0.13816959857940675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,8,1,power_law_1.01,0.3808511972427368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,8,1,power_law_1.2,0.1831168055534363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,8,1,power_law_1.01,0.4955264091491699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,8,1,power_law_1.2,0.2685247898101807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,8,1,power_law_1.01,0.5595456123352051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,8,1,power_law_1.2,0.3569855928421021
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,8,1,power_law_1.01,0.7027135848999023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,8,1,power_law_1.2,0.3681663990020752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,8,1,power_law_1.2,0.37207679748535155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,8,1,power_law_1.01,0.9758975982666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,8,1,power_law_1.2,0.3796031951904297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,8,1,power_law_1.2,0.37830400466918945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,8,1,power_law_1.01,1.352025604248047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,8,1,power_law_1.2,0.39378559589385986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,8,1,power_law_1.2,0.029126399755477907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,8,1,power_law_1.2,0.43013758659362794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,8,1,power_law_1.2,0.03351680040359497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,8,1,power_law_1.2,0.43479042053222655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,8,1,power_law_1.2,0.05441280007362366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,8,1,power_law_1.01,0.05263360142707825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,8,1,power_law_1.2,0.07862399816513062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,8,1,balanced,0.09850666920344035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,8,1,power_law_1.2,0.10725760459899902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,8,1,power_law_1.2,0.14528640508651733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,8,1,power_law_1.2,0.22935678958892822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,8,1,power_law_1.2,0.2395711898803711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,8,1,power_law_1.2,0.24963839054107667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,8,1,power_law_1.2,0.2529920101165771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,8,1,power_law_1.2,0.4524672031402588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,8,1,power_law_1.2,0.2664383888244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,8,1,power_law_1.2,0.4671807765960693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,8,1,power_law_1.2,0.2657407999038696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,8,1,power_law_1.2,0.48662400245666504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,8,1,power_law_1.2,0.2834111928939819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,8,1,power_law_1.2,0.6199168205261231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,8,1,power_law_1.2,0.29717121124267576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,8,1,power_law_1.2,0.30585598945617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,8,1,power_law_1.2,0.6491519927978515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,8,1,power_law_1.2,0.3194240093231201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,8,1,power_law_1.2,0.7558911800384521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,8,1,power_law_1.2,0.3339904069900513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,8,1,power_law_1.2,0.4507775783538818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,8,1,power_law_1.2,0.8130623817443847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,8,1,power_law_1.2,0.48063998222351073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,8,1,power_law_1.2,0.9661760330200195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,8,1,power_law_1.2,0.5479936122894287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,8,1,power_law_1.2,1.1243136405944825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,8,1,power_law_1.2,0.6101503849029541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,8,1,power_law_1.2,0.7390143871307373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,8,1,power_law_1.2,1.4265088081359862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,8,power_law_1.01,0.3995455980300903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,8,1,power_law_1.2,0.8725567817687988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,8,power_law_1.01,0.4106624126434326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,8,power_law_1.01,0.4193280220031738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,8,1,power_law_1.2,1.7375808715820313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,8,1,power_law_1.2,1.126918411254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,8,power_law_1.01,0.4370880126953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,8,1,power_law_1.2,1.3907456398010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,8,1,power_law_1.2,2.393939208984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,8,1,power_law_1.2,1.9231039047241212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,8,balanced,0.5302826563517252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,8,1,power_law_1.2,2.3771520614624024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,8,balanced,0.5766293207804362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,8,1,power_law_1.2,3.0356800079345705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,8,1,power_law_1.2,2.4611135482788087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,8,1,power_law_1.2,3.0174720764160154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,8,1,power_law_1.01,0.07640960216522216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,8,1,power_law_1.2,4.1630207061767575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,8,1,balanced,0.184879998366038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,8,power_law_1.2,0.40268797874450685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,8,1,power_law_1.2,5.354220962524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,8,power_law_1.2,0.4017471790313721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,8,power_law_1.2,0.412934398651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,8,power_law_1.2,0.424070405960083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,8,1,power_law_1.2,0.025824001431465148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,8,1,power_law_1.2,0.027878400683403016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,8,1,power_law_1.2,0.03720319867134094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,8,1,power_law_1.2,0.038464000821113585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,8,1,power_law_1.2,0.051545602083206174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,8,1,power_law_1.2,0.06795520186424256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,8,1,power_law_1.2,0.09348480105400085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,8,1,power_law_1.2,7.720819091796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,8,1,power_law_1.2,0.11338880062103271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,8,1,power_law_1.2,0.12218879461288452
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,8,1,power_law_1.2,0.12699520587921143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,8,1,power_law_1.2,0.1295359969139099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,8,1,power_law_1.2,0.13786879777908326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,8,1,power_law_1.2,0.145414400100708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,8,1,power_law_1.2,0.15368959903717042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,8,1,power_law_1.2,0.16926079988479614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,8,1,power_law_1.2,0.18268799781799316
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,8,1,power_law_1.2,0.1974720001220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,8,1,power_law_1.2,0.22815999984741211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,8,1,power_law_1.2,10.110681915283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,8,1,power_law_1.2,0.22366080284118653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,8,balanced,0.02777066578467687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,8,balanced,0.027461332579453785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,8,balanced,0.029578665892283123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,8,balanced,0.03551466763019562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,8,1,power_law_1.2,0.26480000019073485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,8,balanced,0.05283733208974203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,8,balanced,0.05602666735649109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,8,balanced,0.05874133110046387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,8,balanced,0.0580320010582606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,8,balanced,0.06217066446940104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,8,1,power_law_1.2,0.27857921123504636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,8,balanced,0.06291733185450236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,8,balanced,0.06419200201829274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,8,balanced,0.06727466483910878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,8,balanced,0.06822933256626129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,8,balanced,0.07060799996058147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,8,balanced,0.07458133498827617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,8,balanced,0.0784800002972285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,8,balanced,0.08481599887212117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,8,balanced,0.10075199604034424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,8,balanced,0.11157866319020589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,8,balanced,0.13597333431243896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,8,1,power_law_1.2,0.052051198482513425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,8,balanced,0.161381334066391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,8,balanced,0.2257279952367147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,8,balanced,0.27244265874226886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,8,balanced,0.3821066617965698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,8,balanced,0.4893226623535156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,8,1,power_law_1.01,0.09880319833755494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,8,balanced,0.7067146301269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,8,balanced,0.9275946617126465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,8,balanced,0.027306665976842243
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,8,1,power_law_1.2,0.3316031932830811
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,8,1,power_law_1.2,0.3846719980239868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,8,1,power_law_1.2,0.5041471958160401
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,8,1,power_law_1.2,0.5723264217376709
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,8,1,power_law_1.2,0.7114048004150391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,8,balanced,0.02771199991305669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,8,power_law_1.01,0.4656703948974609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,8,balanced,0.029450667401154835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,8,balanced,0.03148266673088074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,8,balanced,0.03982933362325033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,8,balanced,0.04142933338880539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,8,1,power_law_1.2,0.0769536018371582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,8,1,power_law_1.2,0.980345630645752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,8,balanced,0.050069332122802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,8,power_law_1.01,0.4719103813171387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,8,balanced,0.04977599779764811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,8,balanced,0.047925333182017006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,8,balanced,0.04994666576385498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,8,balanced,0.05207466582457224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,8,power_law_1.01,0.4925824165344238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,8,balanced,0.05402666827042898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,8,balanced,0.05612266560395559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,8,balanced,0.6180320183436075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,8,balanced,0.05827199916044871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,8,balanced,0.0642986645301183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,8,1,power_law_1.2,1.3643327713012696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,8,balanced,0.07038400073846181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,8,balanced,0.07859200239181519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,8,balanced,0.036415999134381614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,8,balanced,0.09701333443323772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,8,balanced,0.035274667044480644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,8,balanced,0.10974933703740437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,8,balanced,0.037471999724706016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,8,balanced,0.13796266913414001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,8,balanced,0.16481066743532816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,8,balanced,0.23215999205907187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,8,balanced,0.2857973376909892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,8,balanced,0.40219199657440186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,8,balanced,0.5174026489257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,8,balanced,0.749130646387736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,8,balanced,0.9825173219045004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,8,balanced,0.03803733239571253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,8,balanced,0.04587733248869578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,8,balanced,0.05676266551017761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,8,balanced,0.05611200133959452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,8,balanced,0.056101332108179726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,8,balanced,0.06637866795063019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,8,balanced,0.0662613312403361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,8,balanced,0.06725866595904033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,8,balanced,0.07054933408896129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,8,balanced,0.07100266714890797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,8,balanced,0.07446933289368947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,8,balanced,0.08053866525491078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,8,balanced,0.08763200044631958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,8,balanced,0.0990559955437978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,8,balanced,0.11673067013422649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,8,balanced,0.1337493360042572
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,8,balanced,0.023904000719388325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,8,balanced,0.19144533077875772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,8,balanced,0.025455998877684276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,8,balanced,0.025792000194390614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,8,balanced,0.027456000447273254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,8,balanced,0.030165334542592365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,8,balanced,0.03151999910672506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,8,balanced,0.03309333324432373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,8,balanced,0.03364799916744232
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,8,balanced,0.03552533437808355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,8,balanced,0.03570133447647095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,8,balanced,0.03587199995915095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,8,balanced,0.04379733403523763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,8,balanced,0.043141335248947144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,8,1,balanced,0.30534933010737103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,8,balanced,0.03862933317820231
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,8,balanced,0.06443200012048085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,8,1,balanced,0.30819199482599896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,8,balanced,0.05885333319505056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,8,balanced,0.04861866434415182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,8,1,power_law_1.2,0.09547520279884339
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,8,balanced,0.04990933338801066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,8,balanced,0.0506933331489563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,8,balanced,0.05605333546797434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,8,balanced,0.06065066655476888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,8,balanced,0.0867733359336853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,8,balanced,0.10325866937637329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,8,balanced,0.1381173332532247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,8,balanced,0.23004267613093057
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,8,balanced,0.17307200034459433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,8,balanced,0.3206239938735962
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,8,balanced,0.2424266735712687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,8,balanced,0.41012799739837646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,8,balanced,0.31752000252405804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,8,1,power_law_1.01,0.13216639757156373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,8,balanced,0.5887840191523234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,8,power_law_1.01,0.031219199299812317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,8,balanced,0.7705866495768229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,8,power_law_1.01,0.04287999868392944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,8,power_law_1.01,0.04496639966964722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,8,power_law_1.01,0.04490880072116852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,8,power_law_1.01,0.04395520091056824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,8,power_law_1.01,0.05190399885177612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,8,power_law_1.01,0.05381119847297668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,8,power_law_1.01,0.05560320019721985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,8,power_law_1.01,0.060678398609161376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,8,power_law_1.01,0.05991680026054382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,8,power_law_1.01,0.06293119788169861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,8,power_law_1.01,0.06480000019073487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,8,balanced,1.1216639677683513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,8,power_law_1.01,0.0655680000782013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,8,power_law_1.01,0.07006080150604248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,8,balanced,1.480826695760091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,8,power_law_1.01,0.07418879866600037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,8,power_law_1.01,0.07921280264854431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,8,power_law_1.01,0.029702401161193846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,8,power_law_1.01,0.08734719753265381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,8,power_law_1.01,0.034643200039863584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,8,power_law_1.01,0.10617599487304688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,8,power_law_1.01,0.03270399868488312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,8,power_law_1.01,0.12703360319137574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,8,power_law_1.01,0.034457600116729735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,8,power_law_1.01,0.15642240047454833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,8,power_law_1.01,0.03792639970779419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,8,power_law_1.01,0.18496639728546144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,8,power_law_1.01,0.040575999021530154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,8,power_law_1.01,0.04593279957771301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,8,power_law_1.01,0.24486401081085205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,8,1,power_law_1.2,0.12519680261611937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,8,power_law_1.01,0.04792320132255554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,8,power_law_1.01,0.30478079319000245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,8,power_law_1.01,0.04945279955863953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,8,power_law_1.01,0.049158400297164916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,8,power_law_1.2,0.46277761459350586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,8,power_law_1.01,0.44181118011474607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,8,power_law_1.01,0.04896639883518219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,8,power_law_1.2,0.4822976112365723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,8,power_law_1.01,0.05324800014495849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,8,power_law_1.01,0.5643072128295898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,8,power_law_1.01,0.05440639853477478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,8,power_law_1.2,0.5025472164154052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,8,power_law_1.01,0.060844802856445314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,8,power_law_1.01,0.8187071800231933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,8,power_law_1.01,0.0663424015045166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,8,power_law_1.01,0.07251840233802795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,8,power_law_1.01,0.0821951985359192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,8,power_law_1.01,1.0402112007141113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,8,power_law_1.01,0.10012799501419067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,8,power_law_1.01,0.0350271999835968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,8,power_law_1.01,0.11708799600601197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,8,power_law_1.01,0.03800959885120392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,8,power_law_1.01,0.149452805519104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,8,power_law_1.01,0.04239999949932098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,8,power_law_1.01,0.18346879482269288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,8,power_law_1.01,0.04357120096683502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,8,power_law_1.01,0.24744958877563478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,8,power_law_1.01,0.0445248007774353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,8,power_law_1.01,0.30290560722351073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,8,power_law_1.01,0.04864639937877655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,8,power_law_1.01,0.05120000243186951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,8,power_law_1.01,0.4307263851165771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,8,power_law_1.01,0.05087360143661499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,8,power_law_1.01,0.54999680519104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,8,power_law_1.01,0.06544640064239501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,8,power_law_1.01,0.0646016001701355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,8,power_law_1.01,0.8162943840026855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,8,power_law_1.01,0.06558719873428345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,8,power_law_1.01,0.07024000287055969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,8,power_law_1.01,0.0706496000289917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,8,power_law_1.01,1.06627197265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,8,power_law_1.01,0.07932159900665284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,8,power_law_1.01,0.08431360125541687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,8,power_law_1.01,0.0917568027973175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,8,power_law_1.01,0.11196160316467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,8,balanced,0.7297493616739908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,8,power_law_1.01,0.13625600337982177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,8,balanced,0.8006560007731119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,8,power_law_1.01,0.1554111957550049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,8,power_law_1.01,0.5414720058441163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,8,power_law_1.01,0.20353920459747316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,8,power_law_1.01,0.2636672019958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,8,power_law_1.01,0.02341119945049286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,8,power_law_1.01,0.6377535820007324
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,8,power_law_1.01,0.02563839852809906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,8,power_law_1.01,0.36487040519714353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,8,power_law_1.01,0.025497600436210632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,8,power_law_1.01,0.45453438758850095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,8,power_law_1.01,0.026694399118423463
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,8,power_law_1.01,0.026822400093078614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,8,power_law_1.01,0.6536640167236328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,8,1,power_law_1.2,0.15187840461730956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,8,power_law_1.01,0.0290367990732193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,8,power_law_1.01,0.030380800366401672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,8,power_law_1.01,0.8530879974365234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,8,1,power_law_1.2,0.20621440410614014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,8,power_law_1.01,0.03173120021820068
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,8,power_law_1.01,0.03327359855175018
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,8,power_law_1.01,0.033504000306129454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,8,power_law_1.01,1.2776384353637695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,8,power_law_1.01,0.03455359935760498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,8,power_law_1.01,0.038726401329040525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,8,1,balanced,0.30606400966644287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,8,power_law_1.01,0.03821440041065216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,8,1,balanced,0.31085866689682007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,8,power_law_1.01,1.711680030822754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,8,power_law_1.01,0.03607679903507233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,8,power_law_1.2,0.03107840120792389
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,8,power_law_1.01,0.05124480128288269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,8,power_law_1.2,0.03550719916820526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,8,power_law_1.01,0.04912639856338501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,8,power_law_1.2,0.03938559889793396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,8,power_law_1.01,0.04712960124015808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,8,1,power_law_1.01,0.16784000396728516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,8,power_law_1.2,0.0436928004026413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,8,1,power_law_1.01,0.21071999073028563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,8,power_law_1.2,0.04378879964351654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,8,power_law_1.2,0.05020800232887268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,8,power_law_1.2,0.05242239832878113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,8,power_law_1.2,0.05562880039215088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,8,power_law_1.2,0.06048640012741089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,8,power_law_1.2,0.06090239882469177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,8,power_law_1.2,0.06288639903068542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,8,power_law_1.2,0.0657151997089386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,8,power_law_1.2,0.06720640063285828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,8,power_law_1.01,0.050886398553848265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,8,power_law_1.2,0.07039999961853027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,8,power_law_1.01,0.054790401458740236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,8,power_law_1.2,0.07485439777374267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,8,power_law_1.01,0.06484479904174804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,8,power_law_1.2,0.07902079820632935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,8,power_law_1.01,0.07256320118904114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,8,power_law_1.2,0.089683198928833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,8,power_law_1.01,0.09749119877815246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,8,power_law_1.2,0.109990394115448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,8,power_law_1.2,0.13121919631958007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,8,power_law_1.2,0.15715839862823486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,8,power_law_1.01,0.12285439968109131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,8,power_law_1.01,0.1621440052986145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,8,power_law_1.2,0.1923967957496643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,8,1,power_law_1.2,0.21841919422149658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,8,power_law_1.2,0.25375359058380126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,8,1,power_law_1.2,0.22866559028625488
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,8,power_law_1.01,0.2129983901977539
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,8,power_law_1.01,0.28641281127929685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,8,1,balanced,0.28697067499160767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,8,1,balanced,0.2911253372828166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,8,power_law_1.2,0.561030387878418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,8,1,balanced,0.29414933919906616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,8,power_law_1.01,0.36785919666290284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,8,1,balanced,0.29755733410517377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,8,1,balanced,0.3055466612180074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,8,power_law_1.2,0.029721599817276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,8,power_law_1.2,0.6489984035491944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,8,1,balanced,0.3119306763013204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,8,power_law_1.2,0.030118399858474733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,8,power_law_1.2,0.03529599905014038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,8,power_law_1.2,0.035385599732398985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,8,power_law_1.2,0.03570559918880463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,8,power_law_1.2,0.039078399538993835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,8,power_law_1.2,0.047603198885917665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,8,power_law_1.2,0.3083391904830933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,8,power_law_1.2,0.046982398629188536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,8,power_law_1.2,0.04790399968624115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,8,power_law_1.2,0.04839679896831513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,8,power_law_1.2,0.048582398891448976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,8,1,balanced,0.3193119963010152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,8,1,power_law_1.2,0.24412798881530762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,8,1,balanced,0.3425706624984741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,8,power_law_1.2,0.055180799961090085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,8,1,balanced,0.4430933396021525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,8,1,power_law_1.2,0.251859188079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,8,power_law_1.2,0.05639680027961731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,8,1,balanced,0.4649386803309123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,8,power_law_1.2,0.061952000856399535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,8,1,power_law_1.2,0.2699712038040161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,8,1,balanced,0.6616586844126383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,8,power_law_1.2,0.06706560254096985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,8,1,power_law_1.2,0.2785279989242554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,8,power_law_1.2,0.0742464005947113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,8,power_law_1.2,0.08388479948043823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,8,1,power_law_1.2,0.2972415924072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,8,power_law_1.2,0.10242559909820556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,8,1,power_law_1.2,0.3321727991104126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,8,power_law_1.2,0.12140799760818481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,8,power_law_1.2,0.4525311946868896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,8,1,power_law_1.2,0.3567296028137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,8,power_law_1.2,0.14825600385665894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,8,power_law_1.2,0.19100159406661987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,8,1,power_law_1.2,0.3624576091766357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,8,power_law_1.2,0.24892799854278563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,8,1,power_law_1.2,0.4654975891113281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,8,balanced,1.2268319924672444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,8,1,power_law_1.2,0.5724160194396972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,8,1,balanced,0.8487892945607504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,8,balanced,1.4086772600809734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,8,1,power_law_1.2,0.7600512027740478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,8,power_law_1.01,0.697433614730835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,8,balanced,2.017418702443441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,8,1,power_law_1.2,0.9608768463134766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,8,balanced,2.60591459274292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,8,1,power_law_1.01,0.22648959159851073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,8,power_law_1.2,0.31189761161804197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,8,1,power_law_1.01,0.22433919906616212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,8,balanced,3.762949307759603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,8,balanced,4.894821484883626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,8,power_law_1.2,0.03480960130691528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,8,power_law_1.2,0.03640959858894348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,8,power_law_1.2,0.040940800309181215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,8,power_law_1.2,0.04227840006351471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,8,power_law_1.2,0.043398401141166686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,8,power_law_1.2,0.05010560154914856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,8,power_law_1.2,0.04945279955863953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,8,power_law_1.2,0.054662400484085084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,8,power_law_1.2,0.06458240151405334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,8,power_law_1.2,0.06555520296096802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,8,power_law_1.2,0.06718720197677612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,8,power_law_1.2,0.06840959787368775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,8,power_law_1.2,0.07246720194816589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,8,power_law_1.2,0.076665598154068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,8,power_law_1.2,0.08314239978790283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,8,power_law_1.2,0.09494400024414062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,8,power_law_1.2,0.11128959655761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,8,power_law_1.2,0.13727999925613404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,8,power_law_1.2,0.162118399143219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,8,power_law_1.2,0.21714560985565184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,8,power_law_1.2,0.271232008934021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,8,power_law_1.2,0.37890560626983644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,8,power_law_1.2,0.5806464195251465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,8,power_law_1.2,0.5014976024627685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,8,1,power_law_1.01,0.24185600280761718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,8,1,power_law_1.01,0.2543936014175415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,8,power_law_1.2,0.4382656097412109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,8,1,power_law_1.01,0.2675391912460327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,8,1,power_law_1.01,0.27333118915557864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,8,1,power_law_1.2,2.054329681396484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,8,1,power_law_1.01,0.29266560077667236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,8,1,power_law_1.01,0.3302783966064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,8,1,power_law_1.01,0.35092480182647706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,8,1,power_law_1.2,2.594918441772461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,8,1,power_law_1.01,0.35571839809417727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,8,1,power_law_1.01,0.461740779876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,8,1,power_law_1.01,0.5477824211120605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,8,1,power_law_1.2,3.5705791473388673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,8,1,power_law_1.01,0.752019214630127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,8,power_law_1.2,0.5667967796325684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,8,power_law_1.2,0.6977983951568604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,8,1,power_law_1.01,0.9416255950927734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,8,power_law_1.2,0.7507328033447266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,8,1,power_law_1.2,4.595072174072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,8,1,power_law_1.2,6.642918395996094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,8,power_law_1.01,0.9012607574462891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,8,power_law_1.01,1.109894371032715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,8,1,balanced,1.8035786946614583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,8,1,balanced,2.2329066594441733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,8,1,balanced,3.3656479517618814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,8,1,power_law_1.2,8.685990142822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,8,1,balanced,4.401482582092285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,8,power_law_1.2,0.02542720139026642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,8,1,balanced,6.490730921427409
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,8,power_law_1.2,0.026150399446487428
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,8,power_law_1.2,0.027385601401329042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,8,power_law_1.2,0.028217598795890808
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,8,power_law_1.2,0.028038400411605834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,8,power_law_1.2,0.030291199684143066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,8,power_law_1.2,0.031814399361610415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,8,power_law_1.2,0.03357439935207367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,8,power_law_1.2,0.03492479920387268
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,8,power_law_1.2,0.034867200255393985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,8,1,balanced,8.556922912597656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,8,power_law_1.2,0.03562879860401154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,8,balanced,0.03568533311287562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,8,balanced,0.03977066775163015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,8,power_law_1.2,0.04059520065784454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,8,balanced,0.055946667989095054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,8,balanced,0.08076799909273784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,8,balanced,0.12549333771069845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,8,power_law_1.2,0.0381632000207901
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,8,power_law_1.2,0.03624320030212402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,8,power_law_1.2,0.050432002544403075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,8,power_law_1.2,0.04947839975357056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,8,power_law_1.2,0.049107199907302855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,8,power_law_1.2,0.05377920269966126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,8,power_law_1.2,0.05904639959335327
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,8,power_law_1.2,0.06636160016059875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,8,power_law_1.2,0.07767680287361145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,8,balanced,0.12829333543777466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,8,balanced,0.14882666865984598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,8,balanced,0.1486293375492096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,8,balanced,0.13935466607411703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,8,balanced,0.1405333379904429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,8,balanced,0.150629331668218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,8,power_law_1.2,0.875276756286621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,8,balanced,0.15440533558527628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,8,balanced,0.15923733512560526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,8,balanced,0.16422933340072632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,8,balanced,0.17058134078979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,8,power_law_1.2,0.10808320045471191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,8,balanced,0.17838933070500693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,8,balanced,0.19173866510391235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,8,balanced,0.23457600673039755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,8,balanced,0.25507734219233197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,8,balanced,0.292303999265035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,8,balanced,0.33033066987991333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,8,balanced,0.47310932477315265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,8,balanced,0.5556853214899699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,8,balanced,0.7901279926300049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,8,balanced,1.012991984685262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,8,balanced,1.4807039896647136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,8,power_law_1.2,0.12188160419464111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,8,power_law_1.2,0.9539199829101562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,8,balanced,1.9336585998535156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,8,balanced,0.0334346666932106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,8,balanced,0.03575466573238373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,8,balanced,0.041797334949175514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,8,power_law_1.2,1.1625215530395507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,8,balanced,0.062021334966023765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,8,balanced,0.08654399712880452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,8,balanced,0.08755733569463094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,8,balanced,0.10251733660697937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,8,balanced,0.10290666421254475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,8,balanced,0.10801066954930623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,8,balanced,0.10801600416501363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,8,balanced,0.11352533102035522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,8,balanced,0.11954666177431743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,8,balanced,0.12372799714406331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,8,balanced,0.13012799620628357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,8,1,power_law_1.01,2.039321517944336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,8,balanced,0.13773866494496664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,8,balanced,0.14592533310254416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,8,balanced,0.16238933801651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,8,balanced,0.18921599785486856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,8,balanced,0.21235734224319458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,8,balanced,0.25778132677078247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,8,balanced,0.30241600672403973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,8,balanced,0.4298400084177653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,8,1,power_law_1.01,2.546950340270996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,8,balanced,0.5185866753260294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,8,power_law_1.2,0.8960895538330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,8,balanced,0.7381280263264974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,8,balanced,0.9515840212504069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,8,balanced,1.3791680335998535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,8,1,power_law_1.01,3.567174530029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,8,power_law_1.01,1.4595840454101563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,8,balanced,1.8122560183207195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,8,balanced,0.047728002071380615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,8,balanced,0.047728002071380615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,8,balanced,0.05605866511662801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,8,power_law_1.01,1.772812843322754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,8,balanced,0.07010133564472198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,8,balanced,0.09040533502896626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,8,balanced,0.0937493344148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,8,balanced,0.09404800335566203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,8,balanced,0.095551997423172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,8,balanced,0.15893866618474325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,8,balanced,0.16198399662971497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,8,1,power_law_1.01,4.562201690673828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,8,power_law_1.01,2.579871940612793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,8,power_law_1.01,3.2199039459228516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,8,balanced,0.16473066806793213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,8,balanced,0.17086933056513467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,8,balanced,0.17566933234532675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,8,1,power_law_1.01,6.615596771240234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,8,balanced,0.183514674504598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,8,balanced,0.190938671429952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,8,power_law_1.2,0.16844160556793214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,8,balanced,0.1998293399810791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,8,balanced,0.2182933290799459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,8,balanced,0.24874132871627808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,8,balanced,0.27929067611694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,8,balanced,0.41097064812978107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,8,power_law_1.01,4.660332870483399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,8,balanced,0.4779946804046631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,8,balanced,0.6804160277048746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,8,balanced,0.8797279993693033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,8,balanced,1.2739040056864421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,8,balanced,1.6700053215026855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,8,1,power_law_1.01,8.677568054199218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,8,power_law_1.01,6.302419281005859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,8,balanced,2.4435839653015137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,8,power_law_1.01,0.07596160173416137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,8,power_law_1.01,0.09093760251998902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,8,power_law_1.01,0.10203520059585572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,8,balanced,3.2276268005371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,8,power_law_1.2,0.850169563293457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,8,power_law_1.01,0.1102463960647583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,8,power_law_1.01,0.05548160076141358
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,8,power_law_1.2,0.21439359188079835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,8,power_law_1.01,0.10904959440231324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,8,power_law_1.01,0.06518399715423584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,8,power_law_1.01,0.11806080341339112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,8,power_law_1.01,0.13761279582977295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,8,power_law_1.01,0.13564159870147705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,8,power_law_1.01,0.12796159982681274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,8,power_law_1.01,0.13193600177764891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,8,power_law_1.01,0.14583679437637329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,8,power_law_1.01,0.14711040258407593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,8,power_law_1.01,0.1573632001876831
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,8,balanced,0.028959999481836956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,8,balanced,0.02951466788848241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,8,power_law_1.01,0.17212159633636476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,8,balanced,0.02942933390537898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,8,balanced,0.03342399994532267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,8,power_law_1.01,0.18277759552001954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,8,balanced,0.055888002117474876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,8,balanced,0.0598880002895991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,8,power_law_1.01,0.06406400203704835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,8,balanced,0.060533334811528526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,8,power_law_1.01,0.19652479887008667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,8,balanced,0.06397866706053416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,8,balanced,0.06281599899133046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,8,power_law_1.01,0.07411839962005615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,8,balanced,0.06413333117961884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,8,power_law_1.01,0.20767359733581542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,8,power_law_1.01,0.0704576015472412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,8,power_law_1.01,0.24805119037628173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,8,power_law_1.01,0.08236799836158752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,8,power_law_1.01,0.277728009223938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,8,power_law_1.01,0.09839360117912292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,8,power_law_1.01,0.09678720235824585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,8,power_law_1.01,0.3597503900527954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,8,power_law_1.01,0.0977728009223938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,8,power_law_1.01,0.42269439697265626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,8,power_law_1.01,0.09871360063552856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,8,power_law_1.01,0.11032320261001587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,8,power_law_1.01,0.5762944221496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,8,power_law_1.01,0.11461759805679321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,8,power_law_1.01,0.1220736026763916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,8,power_law_1.01,0.6872000217437744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,8,balanced,0.06398400167624156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,8,power_law_1.01,0.13068799972534179
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,8,balanced,0.08620267113049825
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,8,balanced,0.0788320004940033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,8,power_law_1.01,0.14088319540023803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,8,balanced,0.07049066821734111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,8,power_law_1.01,0.9574015617370606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,8,balanced,0.1711733341217041
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,8,balanced,0.1569386621316274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,8,power_law_1.01,0.1482751965522766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,8,balanced,0.0944160024325053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,8,balanced,0.0848586658636729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,8,power_law_1.01,0.1636672019958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,8,balanced,0.0913706620534261
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,8,balanced,0.10296533505121867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,8,power_law_1.01,1.2113471984863282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,8,balanced,0.11359467109044392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,8,power_law_1.01,0.19610879421234131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,8,balanced,0.15034133195877075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,8,balanced,0.1771413286526998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,8,power_law_1.01,0.23105919361114502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,8,balanced,0.24541334311167398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,8,balanced,0.3185653289159139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,8,power_law_1.01,1.845689582824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,8,balanced,0.4554400046666463
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,8,balanced,0.593397339185079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,8,power_law_1.2,1.5608448028564452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,8,power_law_1.01,0.06566399931907654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,8,power_law_1.2,0.309932804107666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,8,power_law_1.01,2.495084762573242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,8,power_law_1.01,0.0722432017326355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,8,power_law_1.2,1.8884864807128907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,8,power_law_1.01,0.07783039808273315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,8,power_law_1.01,0.0825215995311737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,8,power_law_1.01,0.07550719976425171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,8,power_law_1.01,0.2872704029083252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,8,power_law_1.01,0.08872320055961609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,8,power_law_1.01,0.3498176097869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,8,power_law_1.01,0.08933759927749634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,8,power_law_1.2,2.682566452026367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,8,power_law_1.01,0.09188479781150818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,8,power_law_1.01,0.47133440971374513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,8,power_law_1.01,0.1549888014793396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,8,power_law_1.01,0.15789439678192138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,8,power_law_1.01,0.600435209274292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,8,power_law_1.2,1.088428783416748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,8,power_law_1.01,0.16060800552368165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,8,power_law_1.2,0.07328640222549439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,8,power_law_1.01,0.16702719926834106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,8,power_law_1.01,0.8332287788391113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,8,power_law_1.2,3.393356704711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,8,power_law_1.2,0.08041599988937378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,8,power_law_1.01,0.17223039865493775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,8,power_law_1.2,0.08599680066108703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,8,power_law_1.01,0.18200960159301757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,8,power_law_1.01,1.0484736442565918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,8,power_law_1.2,0.09033600091934205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,8,power_law_1.01,0.1954624056816101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,8,power_law_1.2,0.085452800989151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,8,power_law_1.01,0.20743041038513182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,8,power_law_1.2,0.11775360107421876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,8,power_law_1.01,0.2392512083053589
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,8,power_law_1.01,0.029555198550224305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,8,power_law_1.01,1.5343168258666993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,8,power_law_1.01,0.30243840217590334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,8,power_law_1.01,0.03660799860954285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,8,power_law_1.01,0.04314880073070526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,8,power_law_1.01,0.3667327880859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,8,power_law_1.2,4.940902328491211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,8,power_law_1.01,0.044409599900245664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,8,power_law_1.01,0.47146239280700686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,8,power_law_1.01,2.064518356323242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,8,power_law_1.01,0.043833601474761966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,8,power_law_1.2,0.054176002740859985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,8,power_law_1.01,0.5890560150146484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,8,power_law_1.2,0.056620800495147706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,8,power_law_1.2,0.06648319959640503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,8,power_law_1.01,0.8611264228820801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,8,power_law_1.2,0.07319679856300354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,8,power_law_1.2,0.06883199810981751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,8,power_law_1.01,1.0904767990112305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,8,power_law_1.2,0.08476799726486206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,8,power_law_1.2,0.09867519736289979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,8,power_law_1.2,0.09945600032806397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,8,power_law_1.2,6.863385772705078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,8,power_law_1.01,1.5193023681640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,8,power_law_1.2,0.09771519899368286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,8,power_law_1.01,0.054611200094223024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,8,power_law_1.2,0.10033279657363892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,8,power_law_1.2,0.06440960168838501
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,8,power_law_1.01,0.05697280168533325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,8,power_law_1.2,0.11026560068130493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,8,power_law_1.2,0.0699455976486206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,8,power_law_1.01,0.058745598793029784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,8,power_law_1.2,0.11820160150527954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,8,power_law_1.01,0.06062080264091492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,8,power_law_1.2,0.13790719509124755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,8,power_law_1.2,0.11802879571914673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,8,power_law_1.01,0.05987200140953064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,8,power_law_1.2,0.14169600009918212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,8,power_law_1.2,0.13308800458908082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,8,power_law_1.01,0.06032639741897583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,8,power_law_1.2,0.13185280561447144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,8,power_law_1.2,0.1410367965698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,8,power_law_1.01,0.06930559873580933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,8,power_law_1.2,0.13106559514999389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,8,power_law_1.2,0.1507200002670288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,8,power_law_1.01,0.0679744005203247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,8,power_law_1.2,0.1484287977218628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,8,power_law_1.2,0.16421760320663453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,8,power_law_1.2,0.15194239616394042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,8,power_law_1.2,0.20495998859405518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,8,power_law_1.2,0.15696640014648439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,8,power_law_1.2,0.23774080276489257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,8,power_law_1.2,0.1739967942237854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,8,power_law_1.2,0.18812160491943358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,8,power_law_1.2,0.2953279972076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,8,power_law_1.2,0.07098879814147949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,8,power_law_1.2,0.197817599773407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,8,power_law_1.2,0.36419200897216797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,8,power_law_1.2,0.0803712010383606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,8,power_law_1.2,0.21336960792541504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,8,power_law_1.2,0.07454720139503479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,8,power_law_1.2,0.49183359146118166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,8,power_law_1.2,0.08636800050735474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,8,power_law_1.2,0.08979840278625488
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,8,power_law_1.01,0.06416640281677247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,8,power_law_1.2,0.08932480216026306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,8,power_law_1.2,1.316972827911377
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,8,power_law_1.01,0.13396480083465576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,8,power_law_1.2,0.1541440010070801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,8,power_law_1.2,0.5947455883026123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,8,power_law_1.01,0.11319680213928222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,8,power_law_1.2,0.15872639417648315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,8,power_law_1.01,2.0403392791748045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,8,power_law_1.01,0.08490239977836608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,8,power_law_1.2,0.16316800117492675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,8,power_law_1.01,0.08464000225067139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,8,power_law_1.2,0.1684928059577942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,8,power_law_1.01,0.09600639939308167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,8,power_law_1.2,0.17492480278015138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,8,power_law_1.2,0.25518720149993895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,8,power_law_1.2,0.8927743911743165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,8,power_law_1.01,0.11116160154342651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,8,power_law_1.2,0.18423680067062378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,8,power_law_1.2,0.29142398834228517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,8,power_law_1.01,2.9442752838134765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,8,power_law_1.2,0.19460480213165282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,8,power_law_1.2,0.3601599931716919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,8,power_law_1.2,0.21215999126434326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,8,power_law_1.2,0.4243584156036377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,8,power_law_1.2,0.25956480503082274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,8,power_law_1.2,0.3211647987365723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,8,power_law_1.2,0.5684224128723144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,8,power_law_1.2,0.3772223949432373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,8,power_law_1.2,0.6910016059875488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,8,power_law_1.01,3.800038528442383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,8,power_law_1.2,0.49719038009643557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,8,power_law_1.2,0.6665984153747558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,8,power_law_1.01,0.1345919966697693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,8,power_law_1.01,0.17055360078811646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,8,power_law_1.01,0.21833600997924804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,8,power_law_1.01,0.29490559101104735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,8,power_law_1.01,0.39260799884796144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,8,power_law_1.2,0.8471424102783203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,8,power_law_1.01,0.5402624130249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,8,power_law_1.01,0.8195008277893067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,8,1,balanced,0.03137599925200144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,8,1,balanced,0.03759466608365377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,8,1,balanced,0.05548266569773356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,8,1,balanced,0.0946613351504008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,8,1,balanced,0.16170666615168253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,8,1,balanced,0.16359466314315796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,8,power_law_1.2,1.030022430419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,8,1,balanced,0.17525867621103922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,8,1,balanced,0.17850132783253989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,8,1,balanced,0.14255999525388083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,8,power_law_1.2,0.032569599151611325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,8,1,balanced,0.14357866843541464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,8,1,balanced,0.15266666809717813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,8,1,balanced,0.15076266725858053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,8,power_law_1.2,0.0353408008813858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,8,1,balanced,0.15379732847213745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,8,1,balanced,0.15582933028539023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,8,power_law_1.2,0.04156799912452698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,8,1,balanced,0.16064533591270447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,8,power_law_1.2,1.1419520378112793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,8,1,balanced,0.16471466422080994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,8,1,balanced,0.17356799046198526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,8,power_law_1.2,0.047244799137115476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,8,1,balanced,0.23152534166971842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,8,1,balanced,0.24496533473332724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,8,power_law_1.2,0.04736000001430511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,8,1,balanced,0.26976533730824787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,8,1,balanced,0.29741867383321124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,8,power_law_1.2,0.05422719717025757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,8,1,balanced,0.42204801241556805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,8,power_law_1.2,0.05804799795150757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,8,1,balanced,0.47383999824523926
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,8,power_law_1.2,0.0605184018611908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,8,1,balanced,0.6660000085830688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,8,power_law_1.2,0.05850239992141724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,8,1,balanced,0.8825866381327311
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,8,power_law_1.2,0.06000639796257019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,8,1,balanced,1.2763573328653972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,8,power_law_1.2,0.061260801553726194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,8,power_law_1.2,0.06967039704322815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,8,power_law_1.2,1.6452543258666992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,8,1,balanced,1.683813254038493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,8,1,balanced,0.027162666122118633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,8,1,balanced,0.027509334186712902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,8,1,balanced,0.03766400118668874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,8,1,balanced,0.06397866706053416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,8,1,balanced,0.09715200463930766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,8,1,balanced,0.09780266880989075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,8,1,balanced,0.1165706713994344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,8,1,balanced,0.11634666721026103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,8,power_law_1.2,1.2789376258850098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,8,1,balanced,0.1104853351910909
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,8,1,balanced,0.11183466513951619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,8,1,balanced,0.11555199821790059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,8,1,balanced,0.11728533109029134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,8,1,balanced,0.1200320025285085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,8,1,balanced,0.12444266676902771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,8,1,balanced,0.12988799810409546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,8,1,balanced,0.1320480008920034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,8,1,balanced,0.14013333121935526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,8,1,balanced,0.18995734055836996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,8,1,balanced,0.20510399341583252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,8,1,balanced,0.23156267404556274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,8,1,balanced,0.2613706588745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,8,power_law_1.2,0.06863359808921814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,8,1,balanced,0.3758346637090047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,8,power_law_1.2,0.0658240020275116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,8,1,balanced,0.4342720111211141
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,8,power_law_1.2,0.13787519931793213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,8,1,balanced,0.6076906522115072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,8,power_law_1.2,0.13175679445266725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,8,1,balanced,0.7992586294809977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,8,power_law_1.2,0.08999680280685425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,8,1,balanced,1.1611200173695881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,8,power_law_1.2,0.09366400241851806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,8,power_law_1.2,0.09835519790649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,8,1,balanced,1.5298080444335938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,8,power_law_1.2,0.11573120355606079
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,8,power_law_1.2,0.14325120449066162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,8,power_law_1.2,0.18938239812850952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,8,1,balanced,0.025242666403452556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,8,1,balanced,0.027130665878454845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,8,1,balanced,0.029274667302767437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,8,1,balanced,0.03874133278926214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,8,1,balanced,0.060032000144322716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,8,power_law_1.2,0.22758400440216064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,8,1,balanced,0.06222933530807495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,8,1,balanced,0.06434666613737743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,8,1,balanced,0.06849599877993266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,8,1,balanced,0.0681386689345042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,8,1,balanced,0.07016533116499583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,8,1,balanced,0.07225066423416138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,8,1,balanced,0.08276799817879994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,8,1,balanced,0.08476266264915466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,8,power_law_1.2,0.3053247928619385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,8,1,balanced,0.09099200367927551
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,8,1,balanced,0.0897759993871053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,8,1,balanced,0.09157333771387736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,8,1,balanced,0.0981333355108897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,8,power_law_1.2,1.1087552070617677
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,8,1,balanced,0.10870933532714844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,8,1,balanced,0.1190720001856486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,8,1,power_law_1.01,0.029446399211883544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,8,1,balanced,0.16580800215403238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,8,1,balanced,0.180351992448171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,8,1,power_law_1.01,0.03765760064125061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,8,1,balanced,0.2893279989560445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,8,1,balanced,0.24733332792917886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,8,1,power_law_1.01,0.053478401899337766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,8,1,balanced,0.34939201672871906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,8,1,power_law_1.01,0.07526400089263915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,8,1,balanced,0.44314666589101154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,8,1,power_law_1.01,0.09255040287971497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,8,1,balanced,0.6285866498947144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,8,1,power_law_1.01,0.13877760171890258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,8,power_law_1.2,1.202995204925537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,8,1,power_law_1.01,0.1561087965965271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,8,1,balanced,0.8167200088500977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,8,1,power_law_1.01,0.1564352035522461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,8,1,power_law_1.01,0.02507520020008087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,8,1,power_law_1.01,0.13182079792022705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,8,1,power_law_1.01,0.03028480112552643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,8,1,power_law_1.01,0.13271679878234863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,8,1,power_law_1.01,0.039673599600791934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,8,1,power_law_1.01,0.14362239837646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,8,1,power_law_1.01,0.0548799991607666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,8,1,power_law_1.01,0.14287999868392945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,8,1,power_law_1.01,0.06467840075492859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,8,1,power_law_1.01,0.1477952003479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,8,1,power_law_1.01,0.08480640053749085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,8,1,power_law_1.01,0.1506943941116333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,8,power_law_1.2,0.4061823844909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,8,1,power_law_1.01,0.1049407958984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,8,1,power_law_1.01,0.15741440057754516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,8,1,power_law_1.01,0.10831999778747559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,8,1,power_law_1.01,0.1642688035964966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,8,1,power_law_1.01,0.10484479665756226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,8,1,power_law_1.01,0.19855999946594238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,8,1,power_law_1.01,0.10638079643249512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,8,1,power_law_1.01,0.23417599201202394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,8,1,power_law_1.01,0.1109887957572937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,8,1,power_law_1.01,0.2525887966156006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,8,1,power_law_1.01,0.11344640254974366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,8,1,power_law_1.01,0.2923583984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,8,1,power_law_1.01,0.35733120441436766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,8,power_law_1.2,1.9144384384155273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,8,1,power_law_1.01,0.438431978225708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,8,1,power_law_1.01,0.5445375919342041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,8,1,power_law_1.01,0.7359488010406494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,8,1,power_law_1.01,0.9520000457763672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,8,1,power_law_1.01,0.1168768048286438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,8,1,power_law_1.01,0.12138880491256714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,8,1,power_law_1.01,0.12861440181732178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,8,1,power_law_1.01,1.344921588897705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,8,1,power_law_1.01,0.13276159763336182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,8,1,power_law_1.01,0.1574399948120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,8,1,power_law_1.01,0.19544960260391236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,8,1,power_law_1.01,1.761190414428711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,8,1,power_law_1.01,0.21508479118347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,8,1,power_law_1.01,0.25522561073303224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,8,1,power_law_1.01,0.3082688093185425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,8,1,power_law_1.01,0.38955519199371336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,8,1,power_law_1.01,0.48607358932495115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,8,1,power_law_1.01,0.66211838722229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,8,1,power_law_1.01,0.8526975631713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,8,1,power_law_1.01,0.02393600046634674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,8,1,power_law_1.01,1.2290240287780763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,8,1,power_law_1.01,0.025779199600219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,8,power_law_1.2,0.37985279560089114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,8,1,power_law_1.01,0.029465600848197937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,8,1,power_law_1.01,0.03537920117378235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,8,1,power_law_1.01,1.597913646697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,8,1,power_law_1.01,0.044627198576927186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,8,1,power_law_1.2,0.024940800666809083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,8,1,power_law_1.01,0.05809280276298523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,8,1,power_law_1.2,0.031999999284744264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,8,1,power_law_1.01,0.061267197132110596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,8,1,power_law_1.2,0.03946239948272705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,8,1,power_law_1.01,0.0637440025806427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,8,1,power_law_1.2,0.05148800015449524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,8,1,power_law_1.01,0.06788480281829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,8,1,power_law_1.2,0.06158080101013184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,8,1,power_law_1.01,0.06910079717636108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,8,1,power_law_1.2,0.08440319895744323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,8,1,power_law_1.01,0.0732800006866455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,8,power_law_1.2,2.1620223999023436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,8,1,power_law_1.2,0.10973440408706665
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,8,1,power_law_1.01,0.07777919769287109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,8,1,power_law_1.2,0.11020159721374512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,8,1,power_law_1.2,0.029951998591423036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,8,1,power_law_1.01,0.08335360288619995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,8,1,power_law_1.2,0.10593279600143432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,8,1,power_law_1.2,0.03776000142097473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,8,1,power_law_1.01,0.09216639995574952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,8,1,power_law_1.2,0.10697599649429321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,8,1,power_law_1.2,0.05330560207366943
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,8,1,power_law_1.01,0.08573439717292786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,8,1,power_law_1.2,0.11175680160522461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,8,1,power_law_1.2,0.0710591971874237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,8,1,power_law_1.01,0.09007359743118286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,8,1,power_law_1.2,0.11504640579223632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,8,1,power_law_1.2,0.0893567979335785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,8,1,power_law_1.01,0.10065280199050904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,8,1,power_law_1.2,0.11592320203781128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,8,1,power_law_1.2,0.1429759979248047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,8,1,power_law_1.01,0.10732799768447876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,8,1,power_law_1.2,0.12395520210266113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,8,1,power_law_1.2,0.15763839483261108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,8,1,power_law_1.01,0.12229759693145752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,8,1,power_law_1.2,0.1282431960105896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,8,1,power_law_1.2,0.16225279569625856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,8,1,power_law_1.01,0.1538879990577698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,8,1,power_law_1.2,0.145798397064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,8,1,power_law_1.2,0.13203200101852416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,8,1,power_law_1.01,0.1879040002822876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,8,1,power_law_1.2,0.15970560312271118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,8,1,power_law_1.2,0.13426560163497925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,8,1,power_law_1.2,0.19813120365142822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,8,1,power_law_1.2,0.14216320514678954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,8,1,power_law_1.2,0.21875200271606446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,8,1,power_law_1.2,0.14241280555725097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,8,1,power_law_1.2,0.14577280282974242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,8,1,power_law_1.2,0.2654848098754883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,8,1,power_law_1.2,0.15387519598007202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,8,1,power_law_1.2,0.3134016036987305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,8,1,power_law_1.2,0.15779839754104613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,8,1,power_law_1.2,0.3930624008178711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,8,1,power_law_1.2,0.1731328010559082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,8,1,power_law_1.2,0.19903359413146973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,8,1,power_law_1.2,0.4879039764404297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,8,1,power_law_1.2,0.239302396774292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,8,1,power_law_1.2,0.25540480613708494
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,8,1,power_law_1.01,0.21590399742126465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,8,power_law_1.2,1.5745023727416991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,8,1,power_law_1.2,0.30547199249267576
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,8,1,power_law_1.01,0.2653503894805908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,8,1,power_law_1.2,0.36192638874053956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,8,1,power_law_1.01,0.3608704090118408
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,8,1,power_law_1.01,0.4604800224304199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,8,1,power_law_1.2,0.4429823875427246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,8,1,power_law_1.01,0.6470079898834229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,8,1,power_law_1.2,0.5490880012512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,8,1,power_law_1.01,0.8694144248962402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,8,1,power_law_1.2,0.6659200191497803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,8,1,power_law_1.2,0.024249599874019624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,8,1,power_law_1.2,0.02571519911289215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,8,1,power_law_1.2,0.02943359911441803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,8,1,power_law_1.2,0.034867200255393985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,8,power_law_1.2,2.7044544219970703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,8,balanced,0.04790933430194855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,8,1,power_law_1.2,0.04470399916172028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,8,balanced,0.06381333371003468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,8,balanced,0.0851093331972758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,8,balanced,0.14818666378657022
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,8,1,power_law_1.2,0.05790079832077026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,8,balanced,0.25227200984954834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,8,balanced,0.3137493332227071
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,8,1,power_law_1.2,0.06155520081520081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,8,balanced,0.3304213285446167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,8,1,power_law_1.2,0.06462079882621766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,8,balanced,0.3307466705640157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,8,power_law_1.2,2.2543359756469727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,8,balanced,0.37781866391499835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,8,1,power_law_1.2,0.06929280161857605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,8,balanced,0.03951466580231985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,8,balanced,0.0517546683549881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,8,balanced,0.07009066641330719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,8,1,power_law_1.2,0.06967039704322815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,8,balanced,0.10146666566530864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,8,balanced,0.17071467638015747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,8,balanced,0.19312532742818198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,8,balanced,0.21579732497533163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,8,balanced,0.2182719906171163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,8,balanced,0.22121065855026245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,8,balanced,0.22268799940745035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,8,balanced,0.2480000058809916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,8,balanced,0.2582239905993144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,8,balanced,0.26501333713531494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,8,balanced,0.27533332506815594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,8,balanced,0.28622933228810626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,8,balanced,0.2980426748593648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,8,balanced,0.3183199961980184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,8,balanced,0.37440534432729083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,8,balanced,0.4089866479237874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,8,balanced,0.47979732354482013
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,8,1,power_law_1.2,0.07490559816360473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,8,balanced,0.5489279826482137
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,8,1,power_law_1.2,0.07813760042190551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,8,balanced,0.7836266358693441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,8,1,power_law_1.2,0.08446080088615418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,8,balanced,0.9245440165201823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,8,1,power_law_1.2,0.743232011795044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,8,1,power_law_1.2,0.09214079976081849
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,8,1,power_law_1.2,0.08574720025062561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,8,balanced,1.3077013492584229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,8,1,power_law_1.2,0.09101439714431762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,8,1,power_law_1.2,0.10196479558944702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,8,balanced,1.6752479871114094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,8,balanced,0.3803360064824422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,8,balanced,0.3676266670227051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,8,balanced,0.37717334429423016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,8,power_law_1.2,3.3259456634521483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,8,balanced,0.3816266854604085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,8,balanced,2.3465706507364907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,8,balanced,0.3938239812850952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,8,balanced,0.40197332700093585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,8,balanced,0.4129120111465454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,8,balanced,0.4347466627756755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,8,balanced,0.5089226563771566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,8,balanced,3.1085119247436523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,8,balanced,0.5402079820632935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,8,balanced,0.604693333307902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,8,1,power_law_1.2,0.10780800580978393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,8,1,power_law_1.2,0.12380800247192383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,8,1,power_law_1.2,0.15699199438095093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,8,1,power_law_1.2,0.18736640214920045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,8,1,power_law_1.2,0.2174527883529663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,8,balanced,0.6679626305898031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,8,1,power_law_1.2,0.26765439510345457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,8,balanced,0.9798133373260498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,8,balanced,1.1155146757761638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,8,balanced,1.5821280479431152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,8,1,power_law_1.2,0.8606719970703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,8,balanced,2.128159999847412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,8,1,power_law_1.2,0.9584447860717773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,8,balanced,2.8882932662963867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,8,balanced,3.775946617126465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,8,power_law_1.2,0.6133376121520996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,8,1,power_law_1.2,0.36453120708465575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,8,balanced,0.06005866825580597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,8,balanced,0.030106666187445324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,8,balanced,0.07155199845631917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,8,balanced,0.031632001201311745
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,8,balanced,0.04398933549722036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,8,balanced,0.06318933268388112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,8,balanced,0.09130666653315227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,8,balanced,0.10904533664385478
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,8,balanced,0.1102186640103658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,8,balanced,0.11236799756685893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,8,balanced,0.11371200283368428
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,8,balanced,0.11468266447385152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,8,balanced,0.11557333668073018
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,8,balanced,0.11683199803034465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,8,balanced,0.17218667268753052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,8,balanced,0.15502400199572244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,8,balanced,0.1420906682809194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,8,balanced,0.3275306622187297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,8,balanced,0.25971200068791706
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,8,balanced,0.15173332889874777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,8,balanced,0.15641066431999207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,8,balanced,0.17625067631403604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,8,balanced,0.19562133153279623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,8,balanced,0.273034671942393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,8,balanced,0.3113493323326111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,8,balanced,0.44333867232004803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,8,balanced,0.5740160147349039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,8,balanced,0.7779146830240885
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,8,balanced,1.1445866425832112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,8,power_law_1.01,0.14414080381393432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,8,power_law_1.01,0.22242560386657714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,8,power_law_1.01,0.1855296015739441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,8,1,power_law_1.2,0.46234879493713377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,8,power_law_1.01,0.221343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,8,power_law_1.01,0.21532800197601318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,8,power_law_1.01,0.26944000720977784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,8,power_law_1.01,0.3167871952056885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,8,power_law_1.01,0.30432639122009275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,8,power_law_1.01,0.3646464109420776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,8,power_law_1.01,0.37645440101623534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,8,power_law_1.01,0.34841599464416506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,8,power_law_1.01,0.3431936025619507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,8,power_law_1.01,0.36191999912261963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,8,power_law_1.01,0.38668160438537597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,8,power_law_1.01,0.41911678314208983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,8,power_law_1.01,0.4098368167877197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,8,power_law_1.01,0.4474751949310303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,8,balanced,0.0869706670443217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,8,balanced,0.10692266623179118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,8,power_law_1.01,0.5391104221343994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,8,balanced,0.17835734287897745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,8,balanced,0.21793599923451742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,8,power_law_1.01,0.6055168151855469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,8,power_law_1.01,0.7249599933624268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,8,power_law_1.01,0.8709568023681641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,8,power_law_1.01,1.1193920135498048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,8,1,power_law_1.2,1.349574375152588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,8,power_law_1.01,1.411616039276123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,8,power_law_1.2,1.8076608657836915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,8,power_law_1.01,1.9589632034301758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,8,power_law_1.01,0.09188479781150818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,8,power_law_1.01,0.15354880094528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,8,power_law_1.01,0.14676480293273925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,8,power_law_1.01,2.4719167709350587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,8,power_law_1.01,0.16100480556488037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,8,power_law_1.01,0.1709247946739197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,8,power_law_1.01,0.18366080522537231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,8,power_law_1.01,0.19832320213317872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,8,power_law_1.01,0.199072003364563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,8,power_law_1.01,3.5676734924316404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,8,power_law_1.01,0.21548800468444823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,8,power_law_1.01,0.21875200271606446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,8,power_law_1.01,0.22689919471740722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,8,power_law_1.01,0.25128960609436035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,8,power_law_1.01,0.25170559883117677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,8,power_law_1.01,0.2700416088104248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,8,power_law_1.01,4.519392013549805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,8,power_law_1.01,0.2822144031524658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,8,1,power_law_1.2,1.2288576126098634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,8,power_law_1.01,0.3033792018890381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,8,power_law_1.01,0.3262079954147339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,8,power_law_1.01,0.10085760354995728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,8,power_law_1.01,0.3863552093505859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,8,power_law_1.01,0.1537727952003479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,8,power_law_1.01,0.44330878257751466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,8,power_law_1.01,0.5415167808532715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,8,power_law_1.01,0.6442560195922852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,8,power_law_1.01,0.826534366607666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,8,power_law_1.01,1.0155712127685548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,8,power_law_1.01,1.51461763381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,8,power_law_1.01,1.857689666748047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,8,power_law_1.01,2.7790655136108398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,8,balanced,0.23913600047429404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,8,balanced,0.2450453241666158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,8,balanced,0.24860799312591553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,8,power_law_1.2,4.188627243041992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,8,balanced,0.3477120002110799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,8,balanced,0.35596799850463867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,8,balanced,0.3649066686630249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,8,balanced,0.37253332138061523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,8,balanced,0.38917867342631024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,8,power_law_1.01,3.5116992950439454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,8,balanced,0.4031786521275838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,8,balanced,0.418938676516215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,8,power_law_1.2,0.144652795791626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,8,balanced,0.4455626805623372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,8,power_law_1.2,0.19141119718551636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,8,balanced,0.5007413228352865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,8,power_law_1.2,0.15237760543823242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,8,power_law_1.2,0.21536641120910643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,8,power_law_1.2,0.2758271932601929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,8,power_law_1.2,0.2973311901092529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,8,power_law_1.2,0.315065598487854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,8,power_law_1.2,0.308351993560791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,8,power_law_1.2,0.35116798877716066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,8,power_law_1.2,0.36392960548400877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,8,power_law_1.2,0.35508480072021487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,8,power_law_1.2,0.3600127935409546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,8,power_law_1.01,0.13059840202331544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,8,power_law_1.2,0.37230079174041747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,8,power_law_1.2,0.40944638252258303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,8,power_law_1.2,0.40968961715698243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,8,power_law_1.2,0.41736321449279784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,8,power_law_1.2,0.4847424030303955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,8,power_law_1.01,0.06076800227165222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,8,power_law_1.2,0.5521152019500732
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,8,power_law_1.01,0.08281599879264831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,8,power_law_1.01,0.1577855944633484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,8,power_law_1.2,0.6311615943908692
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,8,power_law_1.01,0.07246720194816589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,8,power_law_1.01,0.18951040506362915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,8,power_law_1.01,0.07927680015563965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,8,power_law_1.2,0.7437888145446777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,8,power_law_1.01,0.19886720180511475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,8,power_law_1.01,0.09270399808883667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,8,power_law_1.01,0.09928320050239563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,8,power_law_1.2,0.9059455871582032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,8,power_law_1.01,0.10122239589691162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,8,power_law_1.01,0.10825599431991577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,8,power_law_1.01,0.10767999887466431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,8,power_law_1.01,0.10769280195236205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,8,power_law_1.01,0.10675200223922729
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,8,power_law_1.01,0.11242239475250244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,8,power_law_1.01,0.13971199989318847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,8,power_law_1.01,0.12700159549713136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,8,power_law_1.01,0.12692480087280272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,8,power_law_1.2,1.1109439849853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,8,power_law_1.01,0.2219264030456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,8,power_law_1.01,0.17166080474853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,8,power_law_1.2,1.4623552322387696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,8,power_law_1.01,0.16363519430160522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,8,power_law_1.01,0.17040640115737915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,8,power_law_1.01,0.20339200496673585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,8,power_law_1.01,0.22960000038146972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,8,power_law_1.01,0.30154879093170167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,8,power_law_1.2,0.8883904457092285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,8,power_law_1.01,0.40785918235778806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,8,power_law_1.01,0.5333504199981689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,8,power_law_1.2,0.09349759817123413
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,8,power_law_1.01,0.6989696025848389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,8,power_law_1.2,0.09945600032806397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,8,power_law_1.2,0.1304128050804138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,8,power_law_1.2,2.0594560623168947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,8,power_law_1.2,0.15441919565200807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,8,power_law_1.01,1.0041664123535157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,8,power_law_1.2,0.1615615963935852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,8,power_law_1.2,0.16962560415267944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,8,1,power_law_1.2,0.6476416110992431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,8,power_law_1.01,1.4783103942871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,8,power_law_1.2,0.19375360012054443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,8,power_law_1.2,0.20410239696502686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,8,power_law_1.2,0.21231999397277831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,8,power_law_1.2,0.22627201080322265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,8,power_law_1.2,0.24107520580291747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,8,power_law_1.2,0.24822399616241456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,8,power_law_1.2,0.24803199768066406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,8,power_law_1.2,2.590163230895996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,8,power_law_1.2,0.2789504051208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,8,power_law_1.2,0.2907072067260742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,8,power_law_1.2,0.3056447982788086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,8,power_law_1.2,0.3339711904525757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,8,power_law_1.2,0.4016704082489014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,8,balanced,0.554757316907247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,8,power_law_1.2,0.45247998237609866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,8,balanced,0.8465387026468912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,8,power_law_1.2,0.5529407978057861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,8,balanced,0.9528640111287435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,8,power_law_1.01,0.21908481121063234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,8,power_law_1.2,0.664799976348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,8,balanced,1.353050708770752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,8,power_law_1.2,0.8755200386047364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,8,balanced,1.737002690633138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,8,1,power_law_1.2,1.77325439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,8,balanced,2.3425280253092446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,8,power_law_1.2,1.109894371032715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,8,balanced,3.0779040654500327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,8,power_law_1.2,1.5922240257263183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,8,power_law_1.2,0.10003199577331542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,8,power_law_1.2,0.11635199785232545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,8,power_law_1.2,1.9816511154174805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,8,balanced,4.43993600209554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,8,power_law_1.2,2.824928092956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,8,balanced,5.8170827229817705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,8,1,balanced,0.044293334086736046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,8,1,balanced,0.06022400160630544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,8,1,balanced,0.09097066521644592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,8,1,balanced,0.16827199856440225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,8,1,balanced,0.29452266295750934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,8,1,balanced,0.3683733145395915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,8,1,balanced,0.37345067660013836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,8,power_law_1.2,3.9145599365234376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,8,1,balanced,0.3749653498331706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,8,1,balanced,0.03129599988460541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,8,1,balanced,0.3808053334554036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,8,power_law_1.01,0.23085439205169678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,8,1,balanced,0.38025065263112384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,8,1,balanced,0.3588693141937256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,8,power_law_1.01,0.22773120403289795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,8,1,balanced,0.35842132568359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,8,1,balanced,0.3598986864089966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,8,power_law_1.01,0.3359231948852539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,8,1,balanced,0.3638613224029541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,8,1,balanced,0.367520014444987
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,8,power_law_1.2,0.05960959792137146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,8,power_law_1.01,0.34079999923706056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,8,1,balanced,0.3747200171152751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,8,1,balanced,0.384117325146993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,8,power_law_1.2,0.06540799736976624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,8,power_law_1.01,0.3498176097869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,8,1,balanced,0.4841066598892212
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,8,power_law_1.2,0.07740160226821899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,8,1,balanced,0.5034666856129965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,8,power_law_1.2,3.8021312713623048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,8,1,balanced,0.5406879981358846
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,8,power_law_1.2,0.08854399919509888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,8,1,balanced,0.046015997727712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,8,1,balanced,0.5791893402735392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,8,1,balanced,0.06532800197601318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,8,1,balanced,0.10297066966692607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,8,power_law_1.2,0.09054719805717468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,8,1,balanced,0.8958986600240072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,8,power_law_1.2,0.09586560130119323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,8,1,balanced,0.9814826647440592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,8,power_law_1.2,0.0982208013534546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,8,power_law_1.2,0.10682239532470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,8,1,balanced,1.4077866872151692
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,8,power_law_1.2,0.10488959550857543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,8,power_law_1.2,0.11015039682388306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,8,power_law_1.01,0.36726400852203367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,8,1,balanced,1.7777706782023113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,8,power_law_1.2,0.10846079587936401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,8,power_law_1.01,0.38815999031066895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,8,power_law_1.2,0.11459840536117553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,8,1,balanced,0.1849600076675415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,8,power_law_1.01,0.4012288093566895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,8,1,balanced,2.3488853772481284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,8,1,balanced,0.21219199895858765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,8,power_law_1.2,0.132915198802948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,8,1,balanced,0.24869332710901895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,8,1,balanced,0.25009065866470337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,8,1,balanced,0.24659732977549234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,8,1,balanced,0.24771199623743692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,8,1,balanced,0.23644800980885824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,8,1,balanced,0.23745065927505493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,8,1,balanced,0.24110400676727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,8,1,balanced,3.2072426478068032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,8,1,balanced,0.24703466892242432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,8,1,balanced,0.2505013346672058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,8,1,balanced,0.2582719922065735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,8,1,balanced,0.2681973377863566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,8,1,balanced,0.3506133159001668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,8,power_law_1.2,0.14981119632720946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,8,1,balanced,0.3708000183105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,8,1,balanced,0.410645325978597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,8,power_law_1.2,0.16308480501174927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,8,1,balanced,0.449285348256429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,8,1,balanced,0.6906507015228271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,8,power_law_1.2,0.14060800075531005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,8,power_law_1.01,0.44202241897583006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,8,power_law_1.2,0.12938239574432372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,8,1,balanced,0.7720106442769369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,8,power_law_1.2,0.19535360336303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,8,power_law_1.01,0.499129581451416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,8,1,balanced,1.1046240329742432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,8,power_law_1.01,0.5899712085723877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,8,1,balanced,1.4109546343485515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,8,power_law_1.01,0.7115456104278565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,8,1,balanced,1.8959360122680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,8,1,power_law_1.2,0.8741439819335938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,8,power_law_1.01,0.9338496208190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,8,1,power_law_1.01,0.04005120098590851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,8,1,balanced,2.536341349283854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,8,1,power_law_1.01,0.0573248028755188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,8,power_law_1.01,1.1837120056152344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,8,1,power_law_1.01,0.08798080086708068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,8,1,power_law_1.01,0.031116798520088196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,8,1,power_law_1.01,0.15605759620666504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,8,1,power_law_1.01,0.04620800018310547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,8,1,balanced,0.02995733420054118
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,8,1,balanced,0.03225066761175791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,8,1,balanced,0.04384533564249674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,8,1,power_law_1.01,0.19066879749298096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,8,1,power_law_1.01,0.06321920156478882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,8,1,balanced,0.06631466746330261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,8,power_law_1.01,1.6076351165771485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,8,1,balanced,0.09641599655151367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,8,1,balanced,0.11370666821797688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,8,1,power_law_1.01,0.2920063972473145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,8,1,balanced,0.11683199803034465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,8,1,balanced,0.12272000312805176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,8,1,balanced,0.12342400352160136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,8,1,balanced,0.12643200159072876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,8,1,balanced,0.13049599528312683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,8,1,balanced,0.1360106666882833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,8,power_law_1.01,2.0459264755249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,8,1,balanced,0.14995200435320535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,8,1,balanced,0.16159466902414957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,8,power_law_1.2,0.13274240493774414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,8,1,balanced,0.17015999555587769
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,8,1,balanced,0.16230400403340658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,8,1,balanced,0.176362673441569
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,8,power_law_1.2,0.19722880125045777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,8,1,balanced,0.18965333700180054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,8,1,balanced,0.20810665686925253
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,8,power_law_1.2,0.19911680221557618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,8,1,balanced,0.30132800340652466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,8,1,balanced,0.30479999383290607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,8,power_law_1.01,3.0803136825561523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,8,power_law_1.2,0.19992320537567138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,8,1,power_law_1.01,0.09127039909362793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,8,1,balanced,0.6442720095316569
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,8,power_law_1.2,0.18101760149002075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,8,1,power_law_1.01,0.12449920177459717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,8,1,power_law_1.01,0.3240511894226074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,8,1,balanced,0.5193013350168864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,8,1,power_law_1.01,0.18166400194168092
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,8,power_law_1.2,0.2203007936477661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,8,1,power_law_1.01,0.34143359661102296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,8,1,balanced,0.7446826299031576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,8,1,power_law_1.01,0.21893761157989503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,8,1,power_law_1.01,0.3574975967407227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,8,power_law_1.2,0.24451839923858643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,8,1,power_law_1.01,0.222489595413208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,8,1,balanced,0.9860693613688151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,8,1,power_law_1.01,0.35925118923187255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,8,1,power_law_1.01,0.22597761154174806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,8,power_law_1.2,0.3540992021560669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,8,1,balanced,1.0410186449686687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,8,1,power_law_1.01,0.33431038856506345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,8,1,power_law_1.01,0.22864000797271727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,8,1,power_law_1.01,0.33539841175079343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,8,1,power_law_1.01,0.2232640027999878
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,8,power_law_1.2,0.41904640197753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,8,1,balanced,1.4585332870483398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,8,1,power_law_1.01,0.2261375904083252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,8,1,power_law_1.01,0.22947840690612792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,8,1,power_law_1.01,0.23992319107055665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,8,1,power_law_1.01,0.24666879177093506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,8,power_law_1.2,0.6146368026733399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,8,1,power_law_1.01,0.2563199996948242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,8,1,power_law_1.01,0.29322240352630613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,8,power_law_1.01,3.940268707275391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,8,1,power_law_1.01,0.3633984088897705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,8,1,power_law_1.01,0.4084671974182129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,8,power_law_1.2,0.7476223945617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,8,1,power_law_1.01,0.4666624069213867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,8,power_law_1.2,1.1120448112487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,8,power_law_1.01,5.596294403076172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,8,1,power_law_1.01,0.3378175973892212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,8,1,power_law_1.01,0.349727988243103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,8,1,power_law_1.01,0.028806400299072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,8,1,power_law_1.01,0.36101760864257815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,8,1,power_law_1.01,0.5410111904144287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,8,1,power_law_1.01,0.030681601166725157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,8,1,power_law_1.01,0.3571135997772217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,8,1,power_law_1.01,0.047065600752830505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,8,1,power_law_1.01,0.686521577835083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,8,1,power_law_1.01,0.38606719970703124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,8,1,power_law_1.01,0.05625600218772888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,8,1,power_law_1.01,0.8302847862243652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,8,power_law_1.01,7.625049591064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,8,1,power_law_1.2,1.610246467590332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,8,1,power_law_1.01,1.10699520111084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,8,power_law_1.2,0.23140480518341064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,8,1,power_law_1.2,0.04021120071411133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,8,1,power_law_1.2,0.030796799063682555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,8,1,power_law_1.2,0.057766401767730714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,8,1,power_law_1.2,0.045535999536514285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,8,1,power_law_1.01,1.401632022857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,8,1,power_law_1.2,0.08755840063095092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,8,1,power_law_1.2,0.06307839751243591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,8,1,power_law_1.2,0.1401087999343872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,8,1,power_law_1.2,0.08447999954223633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,8,1,power_law_1.2,0.17333760261535644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,8,1,power_law_1.2,0.10706559419631959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,8,1,power_law_1.01,0.0716863989830017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,8,1,power_law_1.01,1.9812416076660155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,8,1,power_law_1.2,0.2842623949050903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,8,1,power_law_1.2,0.1825600028038025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,8,1,power_law_1.01,0.09802240133285522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,8,1,power_law_1.2,0.21461119651794433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,8,1,power_law_1.2,0.3326272010803223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,8,1,power_law_1.01,0.1029695987701416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,8,1,power_law_1.2,0.22167680263519288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,8,1,power_law_1.2,0.3386944055557251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,8,1,power_law_1.01,0.11257599592208863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,8,1,power_law_1.2,0.22581760883331298
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,8,1,power_law_1.01,0.1145408034324646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,8,1,power_law_1.2,0.3476288080215454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,8,1,power_law_1.01,2.564729690551758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,8,1,power_law_1.2,0.23054718971252441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,8,1,power_law_1.01,0.11998080015182495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,8,1,power_law_1.2,0.36345601081848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,8,1,power_law_1.2,0.22483201026916505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,8,1,power_law_1.01,0.12451839447021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,8,1,power_law_1.2,0.3384063959121704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,8,1,power_law_1.2,0.22804479598999022
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,8,1,power_law_1.01,0.134060800075531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,8,1,power_law_1.2,0.3410624027252197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,8,1,power_law_1.2,0.2322943925857544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,8,1,power_law_1.01,0.48549761772155764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,8,1,power_law_1.01,0.14228479862213134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,8,1,power_law_1.2,0.34514560699462893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,8,1,power_law_1.2,0.24227840900421144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,8,1,power_law_1.01,0.560038423538208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,8,1,power_law_1.2,0.35341439247131345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,8,1,power_law_1.2,0.2502912044525146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,8,1,power_law_1.01,0.6247039794921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,8,1,power_law_1.2,0.2713792085647583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,8,1,power_law_1.2,0.36209280490875245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,8,1,power_law_1.2,0.29422080516815186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,8,1,power_law_1.2,0.3838399887084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,8,1,power_law_1.01,0.7154880046844483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,8,1,power_law_1.2,0.3847872018814087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,8,1,power_law_1.2,0.422822380065918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,8,1,power_law_1.01,0.8876607894897461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,8,1,power_law_1.2,0.4123072147369385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,8,1,power_law_1.2,0.5407743930816651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,8,1,power_law_1.2,0.4723008155822754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,8,1,power_law_1.2,0.5719552040100098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,8,1,power_law_1.01,1.0592320442199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,8,1,power_law_1.01,0.15918079614639283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,8,1,power_law_1.2,0.5513408184051514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,8,1,power_law_1.01,0.17571840286254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,8,1,power_law_1.2,0.6289663791656495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,8,1,power_law_1.01,0.15614080429077148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,8,1,power_law_1.01,1.4001791954040528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,8,1,power_law_1.2,0.7318399906158447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,8,1,power_law_1.01,0.17713279724121095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,8,1,power_law_1.2,0.6897151947021485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,8,power_law_1.2,0.2351680040359497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,8,1,power_law_1.01,0.18542720079421998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,8,1,power_law_1.2,0.8997056007385253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,8,power_law_1.2,0.21728639602661132
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,8,1,power_law_1.2,0.028620800375938414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,8,1,power_law_1.01,0.21441919803619386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,8,1,power_law_1.2,0.8382719993591309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,8,power_law_1.2,0.3272000074386597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,8,1,power_law_1.2,0.032102400064468385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,8,1,power_law_1.01,0.2532351970672607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,8,power_law_1.2,0.3420480012893677
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,8,1,power_law_1.2,0.043609601259231565
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,8,1,power_law_1.01,0.29985918998718264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,8,1,power_law_1.2,1.1177087783813477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,8,1,power_law_1.2,0.05375999808311462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,8,power_law_1.2,0.36209280490875245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,8,1,power_law_1.2,1.0746432304382325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,8,1,power_law_1.01,0.3593983888626099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,8,1,power_law_1.2,0.07059199810028076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,8,power_law_1.2,0.37173120975494384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,8,1,power_law_1.01,0.4390592098236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,8,1,power_law_1.2,0.09847040176391601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,8,power_law_1.2,0.39255681037902834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,8,1,power_law_1.2,0.10458879470825196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,8,1,power_law_1.2,1.412992000579834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,8,1,power_law_1.01,0.5996799945831299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,8,1,power_law_1.01,1.7568960189819336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,8,1,power_law_1.2,1.4195391654968261
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,8,1,power_law_1.01,0.7655424118041992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,8,1,power_law_1.2,1.9954048156738282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,8,1,power_law_1.01,2.45729923248291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,8,1,power_law_1.2,1.774662399291992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,8,1,power_law_1.01,1.067091178894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,8,1,power_law_1.01,1.4433343887329102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,8,power_law_1.2,0.44142718315124513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,8,balanced,0.05187733471393585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,8,balanced,0.06396799782911937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,8,balanced,0.08990400036176045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,8,1,power_law_1.01,3.2132225036621094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,8,balanced,0.15204800168673197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,8,power_law_1.2,0.4616256237030029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,8,balanced,0.2569493254025777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,8,balanced,0.04192533095677694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,8,balanced,0.4526453415552775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,8,balanced,0.04974400003751119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,8,balanced,0.06899733344713847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,8,balanced,0.09552533427874248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,8,balanced,0.6925973097483317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,8,balanced,0.1539253294467926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,8,balanced,0.27694400151570636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,8,balanced,0.6956533590952555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,8,balanced,0.4376000165939331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,8,balanced,0.7318560282389323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,8,balanced,0.45024001598358154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,8,balanced,0.4190773169199626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,8,balanced,0.7353440125783285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,8,1,power_law_1.2,2.4906688690185548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,8,balanced,0.4227093458175659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,8,balanced,0.8019733428955078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,8,balanced,0.5189386606216431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,8,1,power_law_1.2,0.11163519620895386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,8,balanced,0.5293866793314616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,8,balanced,0.8142346541086832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,8,balanced,0.5358239809672037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,8,1,power_law_1.2,0.11561599969863892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,8,balanced,0.820624033610026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,8,balanced,0.5454133351643881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,8,1,power_law_1.2,0.1194111943244934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,8,balanced,0.8324480056762695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,8,balanced,0.5574506521224976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,8,power_law_1.2,0.5117760181427002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,8,balanced,0.5701386531194051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,8,1,power_law_1.2,0.1270143985748291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,8,balanced,0.5933066606521606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,8,1,power_law_1.2,0.13696639537811278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,8,power_law_1.2,0.6224639892578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,8,balanced,0.6853493054707845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,8,1,power_law_1.2,0.14402559995651246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,8,balanced,0.7200693289438883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,8,1,power_law_1.2,0.16122879981994628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,8,balanced,0.789359966913859
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,8,1,power_law_1.2,0.17713919878005982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,8,balanced,0.8606666723887125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,8,power_law_1.2,1.6579391479492187
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,8,1,power_law_1.2,0.1583232045173645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,8,balanced,1.0
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,8,1,power_law_1.2,0.17955199480056763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,8,balanced,1.1385493278503418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,8,1,power_law_1.2,0.18867199420928954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,8,balanced,0.8432586987813314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,8,1,power_law_1.2,0.21863040924072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,8,balanced,0.8544426759084066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,8,balanced,1.420255978902181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,8,balanced,0.06011733412742615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,8,balanced,0.07447466750939687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,8,balanced,0.8760639826456705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,8,balanced,0.08667199810345967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,8,balanced,0.10804800192515056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,8,balanced,0.1611733337243398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,8,balanced,1.910154660542806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,8,balanced,1.0200479825337727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,8,balanced,0.2829493284225464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,8,balanced,0.40067732334136963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,8,balanced,1.05131729443868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,8,balanced,0.40273598829905194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,8,balanced,0.7289706865946451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,8,balanced,2.477743943532308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,8,balanced,0.7332746982574463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,8,balanced,0.7433546384175619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,8,balanced,0.7499787012736002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,8,balanced,3.2723251978556314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,8,balanced,0.7578933238983154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,8,1,power_law_1.2,0.25656321048736574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,8,balanced,0.7749280134836832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,8,1,power_law_1.2,0.31198720932006835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,8,balanced,0.7916479905446371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,8,power_law_1.2,5.364998245239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,8,1,power_law_1.2,0.37423999309539796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,8,balanced,0.8076266447703043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,8,balanced,1.115594704945882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,8,balanced,0.8350506623586019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,8,power_law_1.01,0.15693440437316894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,8,balanced,0.8883626461029053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,8,balanced,1.1796693007151287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,8,power_law_1.01,0.24879999160766603
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,8,1,power_law_1.2,0.44631037712097166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,8,balanced,0.9440213044484457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,8,balanced,1.308842658996582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,8,1,power_law_1.2,0.6062719821929932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,8,balanced,1.0470613638559978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,8,balanced,1.4412959416707356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,8,balanced,1.1522239844004314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,8,balanced,1.7103625933329265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,8,balanced,1.3623840014139812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,8,balanced,2.3748960494995117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,8,power_law_1.01,0.42960638999938966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,8,balanced,2.9555946985880532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,8,power_law_1.01,0.37960960865020754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,8,power_law_1.01,0.5038976192474365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,8,1,power_law_1.2,2.5851648330688475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,8,power_law_1.01,0.4909503936767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,8,balanced,3.9752480189005532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,8,balanced,1.9835467338562012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,8,1,power_law_1.2,3.2131839752197267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,8,power_law_1.01,0.08811519742012024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,8,balanced,0.029482667644818623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,8,power_law_1.01,0.6158271789550781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,8,balanced,0.03268266717592875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,8,balanced,0.03804266701141993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,8,power_law_1.01,0.14852479696273804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,8,balanced,2.3973546028137207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,8,1,power_law_1.2,0.7676159858703613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,8,power_law_1.01,0.572979211807251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,8,power_law_1.01,0.260479998588562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,8,power_law_1.01,0.10152959823608398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,8,power_law_1.01,0.6326399803161621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,8,power_law_1.01,0.2166975975036621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,8,power_law_1.01,0.15629440546035767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,8,balanced,3.243130683898926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,8,power_law_1.01,0.285696005821228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,8,power_law_1.01,0.6794112205505372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,8,power_law_1.01,0.273305606842041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,8,power_law_1.01,0.3027071952819824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,8,power_law_1.01,0.24924800395965577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,8,power_law_1.01,0.7444863796234131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,8,balanced,0.06217066446940104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,8,balanced,0.09078933795293172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,8,power_law_1.01,0.4038656234741211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,8,balanced,0.14589866995811462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,8,power_law_1.01,0.31166079044342043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,8,balanced,4.5006453196207685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,8,balanced,0.20151466131210327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,8,power_law_1.01,0.7187071800231933
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,8,balanced,0.2055786649386088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,8,balanced,0.2092319925626119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,8,power_law_1.01,0.39711999893188477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,8,power_law_1.01,0.329036808013916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,8,balanced,0.21278399229049683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,8,balanced,0.20773865779240927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,8,power_law_1.01,0.7261375904083252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,8,balanced,0.21150932709376016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,8,power_law_1.01,0.3432703971862793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,8,balanced,0.21793599923451742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,8,power_law_1.01,0.3407615900039673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,8,balanced,0.21758933862050375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,8,balanced,0.22206934293111166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,8,power_law_1.01,0.3368191957473755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,8,balanced,0.2280799945195516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,8,balanced,0.2891146739323934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,8,power_law_1.01,0.3793920040130615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,8,balanced,0.26128532489140827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,8,power_law_1.01,0.4312704086303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,8,1,power_law_1.2,1.0816767692565918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,8,balanced,6.182314554850261
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,8,balanced,0.5738240083058676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,8,balanced,0.353002667427063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,8,power_law_1.01,0.666374397277832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,8,balanced,0.31492799520492554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,8,power_law_1.01,0.47087998390197755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,8,power_law_1.2,0.15433599948883056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,8,balanced,0.35468268394470215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,8,balanced,0.3963786760965983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,8,power_law_1.01,0.6142208099365234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,8,power_law_1.2,0.24786560535430907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,8,power_law_1.01,0.4804671764373779
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,8,balanced,0.4860853354136149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,8,power_law_1.2,0.3492480039596558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,8,balanced,0.6639360189437866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,8,power_law_1.01,0.7999167919158936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,8,balanced,0.8270506858825684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,8,power_law_1.2,0.3617919921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,8,power_law_1.01,0.056435197591781616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,8,power_law_1.01,0.7728576183319091
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,8,balanced,1.238101323445638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,8,power_law_1.2,0.467955207824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,8,power_law_1.01,0.08353279829025269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,8,power_law_1.01,0.7616960048675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,8,power_law_1.2,0.08727679848670959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,8,power_law_1.2,0.5735167980194091
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,8,power_law_1.01,0.13804800510406495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,8,power_law_1.2,0.14833920001983641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,8,power_law_1.01,0.8191871643066406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,8,power_law_1.01,0.12155519723892212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,8,power_law_1.01,0.4710080146789551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,8,power_law_1.2,0.20528640747070312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,8,power_law_1.01,0.9591744422912598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,8,power_law_1.01,0.15733120441436768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,8,power_law_1.01,0.5159423828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,8,power_law_1.01,0.16509439945220947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,8,power_law_1.01,0.9754624366760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,8,power_law_1.01,0.17408640384674073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,8,power_law_1.01,1.1119296073913574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,8,power_law_1.2,0.7606143951416016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,8,power_law_1.2,0.6588287830352784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,8,power_law_1.01,1.1805952072143555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,8,power_law_1.2,0.9924991607666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,8,power_law_1.2,0.21864960193634034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,8,power_law_1.2,0.5662079811096191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,8,power_law_1.01,0.5307136058807373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,8,power_law_1.2,0.27514240741729734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,8,power_law_1.2,1.1727999687194823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,8,power_law_1.2,0.6440703868865967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,8,power_law_1.01,0.5421887874603272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,8,power_law_1.2,0.3575295925140381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,8,power_law_1.2,0.5879231929779053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,8,power_law_1.2,1.6780992507934571
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,8,power_law_1.01,0.1812351942062378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,8,power_law_1.01,0.6309055805206298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,8,power_law_1.2,0.4143551826477051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,8,power_law_1.01,0.6488255977630615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,8,power_law_1.01,0.1863360047340393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,8,power_law_1.2,0.7087744235992431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,8,power_law_1.01,0.6715583801269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,8,power_law_1.2,0.41904640197753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,8,power_law_1.01,0.616761589050293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,8,power_law_1.01,0.1776255965232849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,8,power_law_1.2,2.139219284057617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,8,power_law_1.01,0.7826816082000733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,8,1,power_law_1.2,1.4886143684387207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,8,power_law_1.2,0.7650432109832763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,8,power_law_1.2,0.3996543884277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,8,power_law_1.01,1.4615039825439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,8,power_law_1.01,0.16406400203704835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,8,power_law_1.01,0.6618688106536865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,8,power_law_1.01,0.8820799827575684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,8,power_law_1.2,0.3607615947723389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,8,power_law_1.2,0.7497983932495117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,8,power_law_1.01,0.1778880000114441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,8,power_law_1.01,0.7156032085418701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,8,power_law_1.01,1.6519168853759765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,8,power_law_1.2,0.4525311946868896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,8,power_law_1.01,0.18325120210647583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,8,power_law_1.2,0.6907519817352294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,8,power_law_1.2,3.4833599090576173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,8,power_law_1.01,0.7179776191711426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,8,power_law_1.01,0.1846336007118225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,8,power_law_1.2,0.4850815773010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,8,power_law_1.2,0.7592576026916504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,8,power_law_1.01,0.7175551891326905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,8,power_law_1.01,0.1964735984802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,8,power_law_1.01,2.133951950073242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,8,power_law_1.2,0.4843776226043701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,8,power_law_1.2,0.7669951915740967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,8,power_law_1.01,0.20286080837249756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,8,power_law_1.01,0.7850815773010253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,8,power_law_1.2,0.5016767978668213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,8,power_law_1.01,0.21667840480804443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,8,power_law_1.2,0.7913472175598144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,8,power_law_1.2,0.5349567890167236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,8,power_law_1.01,0.23381760120391845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,8,power_law_1.01,2.6141632080078123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,8,power_law_1.01,1.067142391204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,8,power_law_1.2,0.5150527954101562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,8,power_law_1.2,4.007424163818359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,8,power_law_1.2,0.938265609741211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,8,power_law_1.01,0.3746943950653076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,8,power_law_1.2,0.5677824020385742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,8,power_law_1.01,1.254054355621338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,8,power_law_1.01,0.32646400928497316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,8,power_law_1.2,1.0012160301208497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,8,power_law_1.2,0.6429632186889649
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,8,power_law_1.01,0.31630079746246337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,8,power_law_1.2,1.1266559600830077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,8,power_law_1.01,3.6918785095214846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,8,power_law_1.01,1.6442752838134767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,8,power_law_1.2,0.674124813079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,8,power_law_1.01,0.39164159297943113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,8,power_law_1.01,0.883187198638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,8,power_law_1.2,1.2321536064147949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,8,power_law_1.2,0.7864704132080078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,8,power_law_1.01,0.4356800079345703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,8,power_law_1.01,2.1468671798706054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,8,power_law_1.01,0.968166446685791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,8,power_law_1.01,0.6115392208099365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,8,power_law_1.2,0.881107234954834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,8,power_law_1.2,1.4956095695495606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,8,power_law_1.01,4.697267150878906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,8,power_law_1.01,0.7414591789245606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,8,power_law_1.2,1.1070079803466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,8,power_law_1.2,1.7489856719970702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,8,power_law_1.01,2.765964889526367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,8,power_law_1.2,0.10056320428848267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,8,power_law_1.01,1.0018752098083497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,8,power_law_1.2,1.2948863983154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,8,power_law_1.2,0.15571839809417726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,8,power_law_1.2,0.22960000038146972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,8,power_law_1.2,1.7208768844604492
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,8,power_law_1.01,1.717318344116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,8,power_law_1.2,0.2541248083114624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,8,power_law_1.01,3.5485950469970704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,8,1,balanced,0.043807998299598694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,8,1,balanced,0.05784533421198527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,8,1,balanced,0.08659733335177104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,8,1,balanced,0.17007466157277426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,8,1,balanced,0.03327466547489166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,8,1,balanced,0.04382933179537455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,8,power_law_1.2,0.30538880825042725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,8,1,balanced,0.062218666076660156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,8,1,balanced,0.0916426678498586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,8,1,balanced,0.1691360076268514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,8,power_law_1.2,2.1166080474853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,8,power_law_1.2,0.3741503953933716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,8,power_law_1.2,0.3736576080322266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,8,power_law_1.2,2.280108833312988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,8,power_law_1.2,0.35713920593261717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,8,power_law_1.2,2.988435173034668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,8,power_law_1.01,1.1565119743347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,8,1,balanced,0.3012160062789917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,8,power_law_1.2,0.7066880226135254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,8,1,balanced,0.5511733293533325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,8,power_law_1.2,2.7549440383911135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,8,1,balanced,0.3036800026893616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,8,1,balanced,0.8028746445973715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,8,power_law_1.01,1.3821696281433105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,8,1,balanced,0.5076586802800497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,8,1,balanced,0.8033813635508219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,8,1,balanced,0.5101546843846639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,8,power_law_1.2,0.6937856197357177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,8,1,balanced,0.47354666392008465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,8,1,balanced,0.7337066332499186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,8,1,balanced,0.4759360154469808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,8,power_law_1.2,3.802931213378906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,8,1,balanced,0.7363573710123698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,8,1,balanced,0.5064746538798014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,8,power_law_1.01,1.727769660949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,8,1,balanced,0.784874677658081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,8,1,balanced,0.5058079957962036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,8,power_law_1.2,0.6170176029205322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,8,1,balanced,0.5104320049285889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,8,1,balanced,0.7799946467081705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,8,power_law_1.2,3.833235168457031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,8,1,balanced,0.5176000197728475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,8,power_law_1.2,0.05563520193099976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,8,1,balanced,0.784559965133667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,8,1,balanced,0.5246933301289877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,8,power_law_1.2,0.6852863788604736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,8,1,balanced,0.7893973191579183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,8,1,balanced,0.5325706799825033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,8,power_law_1.2,0.08327680230140685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,8,1,balanced,0.546837329864502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,8,1,balanced,0.8004586696624756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,8,power_law_1.01,2.254425621032715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,8,power_law_1.2,0.11764479875564575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,8,1,balanced,0.7397173245747884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,8,1,balanced,0.8055040041605631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,8,power_law_1.2,0.6188799858093261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,8,1,balanced,0.7604426542917887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,8,1,balanced,0.8192533651987711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,8,power_law_1.2,0.12343679666519165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,8,1,balanced,0.8023200035095215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,8,1,balanced,1.0651733080546062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,8,power_law_1.2,0.6957568168640137
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,8,power_law_1.2,0.14761600494384766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,8,1,balanced,0.8443733056386312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,8,1,balanced,1.0826293627421062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,8,power_law_1.2,5.307936096191407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,8,power_law_1.01,2.9870975494384764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,8,power_law_1.2,0.17961599826812744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,8,power_law_1.2,0.7266687870025634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,8,power_law_1.2,5.938886260986328
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,8,power_law_1.2,0.192576003074646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,8,power_law_1.2,0.7178112030029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,8,power_law_1.2,0.17734400033950806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,8,power_law_1.2,0.7803328037261963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,8,power_law_1.2,0.187225604057312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,8,power_law_1.01,3.7096511840820314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,8,power_law_1.2,0.17099519968032836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,8,1,balanced,0.9371573130289713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,8,power_law_1.2,0.9271552085876464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,8,1,balanced,1.1229013601938884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,8,power_law_1.2,0.16368000507354735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,8,1,balanced,1.0290239651997883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,8,1,balanced,1.1654187043507893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,8,power_law_1.2,0.18884479999542236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,8,1,balanced,1.211130698521932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,8,power_law_1.2,1.0164928436279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,8,1,balanced,1.2533493041992188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,8,power_law_1.2,0.18154879808425903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,8,1,balanced,1.7510933876037598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,8,power_law_1.2,0.1869055986404419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,8,1,balanced,1.345296065012614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,8,power_law_1.01,5.658233642578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,8,power_law_1.2,0.19197440147399902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,8,1,balanced,1.5330133438110352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,8,power_law_1.2,0.2059839963912964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,8,power_law_1.2,0.23455359935760497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,8,1,balanced,2.2384799321492515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,8,power_law_1.2,0.2463167905807495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,8,power_law_1.2,0.3883647918701172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,8,1,balanced,2.626138687133789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,8,power_law_1.2,1.2448575973510743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,8,power_law_1.2,0.3248703956604004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,8,1,balanced,2.122213363647461
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,8,power_law_1.2,0.3698816061019897
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,8,1,balanced,0.03181333343187968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,8,power_law_1.2,1.4203519821166992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,8,1,balanced,0.04784533381462097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,8,1,balanced,0.03249066571394602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,8,power_law_1.01,7.10980453491211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,8,1,balanced,0.03968533376852671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,8,1,balanced,3.5595839818318686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,8,1,balanced,0.06355733176072438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,8,power_law_1.2,0.4106751918792725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,8,1,power_law_1.01,0.04254719913005829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,8,1,balanced,2.8661813735961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,8,1,power_law_1.01,0.033504000306129454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,8,power_law_1.2,0.48686718940734863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,8,1,power_law_1.01,0.056576001644134524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,8,power_law_1.2,1.807436752319336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,8,1,power_law_1.01,0.04164479970932007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,8,1,power_law_1.01,0.0865343987941742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,8,power_law_1.2,0.6303103923797607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,8,1,power_law_1.01,0.06220800280570984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,8,1,power_law_1.01,0.17022080421447755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,8,1,power_law_1.01,0.09101439714431762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,8,power_law_1.2,2.3654144287109373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,8,power_law_1.2,0.7781504154205322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,8,1,power_law_1.01,0.27395200729370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,8,1,power_law_1.01,0.15476479530334472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,8,1,balanced,0.09493866562843323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,8,1,power_law_1.01,0.047328001260757445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,8,1,balanced,0.15530133247375488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,8,1,power_law_1.01,0.3662208080291748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,8,1,power_law_1.01,0.21577599048614501
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,8,1,balanced,0.2174826661745707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,8,power_law_1.2,1.0765055656433105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,8,1,balanced,0.22078933318456015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,8,1,balanced,0.22230400641759238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,8,1,power_law_1.01,0.43625597953796386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,8,1,power_law_1.01,0.28321280479431155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,8,power_law_1.2,3.0982208251953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,8,1,power_law_1.01,0.36187520027160647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,8,power_law_1.2,1.9248512268066407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,8,1,power_law_1.01,0.3479743957519531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,8,1,power_law_1.01,0.3543423891067505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,8,1,power_law_1.01,0.4109248161315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,8,power_law_1.2,4.281190490722656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,8,1,balanced,0.225983997186025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,8,1,balanced,0.23057067394256592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,8,1,power_law_1.01,0.41678719520568847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,8,1,balanced,0.059232001503308616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,8,1,balanced,0.23640533288319907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,8,1,balanced,0.0724426656961441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,8,1,balanced,0.1027733286221822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,8,1,power_law_1.01,0.5645440101623536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,8,1,balanced,0.17464532454808554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,8,1,power_law_1.01,0.424121618270874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,8,1,power_law_1.01,0.521292781829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,8,1,power_law_1.01,0.4360640048980713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,8,1,power_law_1.01,0.5226880073547363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,8,1,power_law_1.01,0.451584005355835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,8,1,power_law_1.01,0.6207104206085206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,8,1,power_law_1.01,0.461516809463501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,8,1,power_law_1.01,0.6273087978363037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,8,1,power_law_1.01,0.47951998710632326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,8,1,balanced,0.2442400058110555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,8,1,balanced,0.2523946762084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,8,1,balanced,0.31917866071065265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,8,1,balanced,0.26255999008814496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,8,1,balanced,0.44863466421763104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,8,1,power_law_1.01,0.6427135944366456
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,8,1,balanced,0.2712053259213765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,8,1,balanced,0.44921600818634033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,8,1,balanced,0.32171199719111127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,8,1,balanced,0.4522560040156047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,8,1,balanced,0.36456533273061115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,8,1,power_law_1.01,0.6584127902984619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,8,1,balanced,0.4550400177637736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,8,1,balanced,0.35354665915171307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,8,1,balanced,0.46668799718221027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,8,1,balanced,0.40932798385620117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,8,1,balanced,0.47073598702748615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,8,1,balanced,0.4293493429819743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,8,1,power_law_1.01,0.670252799987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,8,power_law_1.2,5.46577262878418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,8,1,balanced,0.47229333718617755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,8,1,balanced,0.5158186753590902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,8,1,balanced,0.47175999482472736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,8,1,balanced,0.6998079617818197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,8,1,balanced,0.48131199677785236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,8,1,power_law_1.01,0.6838079929351807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,8,1,balanced,0.48704532782236737
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,8,1,balanced,0.7325279712677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,8,1,balanced,0.5028693278630575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,8,1,power_law_1.01,0.02709760069847107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,8,1,power_law_1.01,0.6539391994476318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,8,1,power_law_1.01,0.7061376094818115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,8,1,balanced,0.5310506820678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,8,1,balanced,1.2063093185424805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,8,1,power_law_1.01,0.02953599989414215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,8,1,power_law_1.01,0.6875904083251954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,8,1,power_law_1.01,0.9243071556091309
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,8,1,balanced,1.0917332967122395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,8,1,power_law_1.01,0.0384768009185791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,8,1,power_law_1.01,0.7736000061035156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,8,1,power_law_1.01,0.8396096229553223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,8,1,power_law_1.01,0.058982402086257935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,8,1,power_law_1.01,0.07167999744415283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,8,1,power_law_1.01,0.10086400508880615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,8,power_law_1.2,7.566732788085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,8,power_law_1.2,8.614144134521485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,8,1,power_law_1.01,0.156550395488739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,8,1,power_law_1.01,0.9908864021301269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,8,1,power_law_1.2,0.043289598822593686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,8,1,power_law_1.2,0.031455999612808226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,8,1,power_law_1.2,0.05655679702758789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,8,1,power_law_1.01,0.9622528076171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,8,1,power_law_1.2,0.041536000370979306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,8,1,balanced,1.5792320569356282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,8,1,power_law_1.01,1.1449343681335449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,8,1,power_law_1.2,0.0865664005279541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,8,1,power_law_1.2,0.06113920211791992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,8,1,power_law_1.01,1.069484806060791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,8,1,power_law_1.2,0.16830079555511473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,8,1,power_law_1.2,0.09049599766731262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,8,1,power_law_1.01,1.4313471794128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,8,1,power_law_1.2,0.2388416051864624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,8,1,power_law_1.2,0.14732799530029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,8,1,power_law_1.01,1.1438336372375488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,8,1,power_law_1.2,0.3346816062927246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,8,1,power_law_1.2,0.20202240943908692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,8,1,power_law_1.01,1.7313983917236329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,8,1,power_law_1.01,1.3111167907714845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,8,1,power_law_1.01,0.06023039817810059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,8,1,power_law_1.2,0.41960959434509276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,8,1,power_law_1.2,0.27048320770263673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,8,1,power_law_1.01,0.08676480054855347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,8,1,power_law_1.2,0.5429632186889648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,8,1,power_law_1.01,1.5134783744812013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,8,1,power_law_1.2,0.35516159534454345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,8,1,power_law_1.01,2.3403903961181642
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,8,1,power_law_1.01,0.11969280242919922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,8,1,power_law_1.2,0.5185472011566162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,8,1,power_law_1.2,0.3387648105621338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,8,1,power_law_1.01,0.14869760274887084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,8,1,power_law_1.01,1.8487167358398438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,8,1,power_law_1.2,0.3441087961196899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,8,1,power_law_1.01,0.17498879432678222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,8,1,power_law_1.01,2.945503997802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,8,1,power_law_1.01,0.1798848032951355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,8,1,power_law_1.2,0.40965118408203127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,8,1,power_law_1.01,2.220185661315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,8,1,power_law_1.01,0.1879744052886963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,8,1,power_law_1.2,0.4192063808441162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,8,1,power_law_1.01,0.19591679573059081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,8,1,power_law_1.2,0.4207744121551514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,8,1,power_law_1.01,0.2037503957748413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,8,1,power_law_1.2,0.4385663986206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,8,1,power_law_1.01,2.962041664123535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,8,1,power_law_1.2,0.5226751804351807
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,8,1,power_law_1.01,0.21480960845947267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,8,1,power_law_1.2,0.4559487819671631
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,8,1,power_law_1.01,0.23124480247497559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,8,1,power_law_1.2,0.6074495792388916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,8,1,power_law_1.2,0.4618368148803711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,8,1,balanced,0.6683839956919352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,8,1,power_law_1.01,0.24597759246826173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,8,1,power_law_1.2,0.6173056125640869
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,8,1,power_law_1.01,0.26458239555358887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,8,1,power_law_1.2,0.5031744003295898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,8,1,power_law_1.01,3.7118785858154295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,8,1,power_law_1.01,0.2975167989730835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,8,1,power_law_1.2,0.6646912097930908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,balanced,0.033626665671666466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,balanced,0.03532800078392029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,balanced,0.04974400003751119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,balanced,0.06348266700903575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,balanced,0.10150399804115295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,8,1,power_law_1.01,0.3764096021652222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,balanced,0.18098666270573935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,8,1,power_law_1.2,0.7041408061981201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,8,1,power_law_1.2,0.04831359982490539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,balanced,0.18826133012771606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,balanced,0.1900213360786438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,balanced,0.21394666035970053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,8,1,power_law_1.01,0.33854079246520996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,balanced,0.21454399824142456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,balanced,0.21177067359288534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,balanced,0.21659733851750693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,8,1,power_law_1.01,0.42853121757507323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,balanced,0.21797333161036173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,balanced,0.22507200638453165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,balanced,0.23292267322540283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,balanced,0.23788267374038696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,8,1,power_law_1.01,0.4194623947143555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,balanced,0.24868800242741904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,balanced,0.29867200056711835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,balanced,0.31483733654022217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,8,1,power_law_1.01,0.5471680164337158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,8,1,power_law_1.2,0.6288064002990723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,balanced,0.35044264793395996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,balanced,0.3866026798884074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,8,1,power_law_1.01,0.5838912010192872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,8,1,power_law_1.2,0.6449984073638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,8,1,power_law_1.2,0.7916287899017334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,8,1,power_law_1.01,0.7806335926055908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,8,1,power_law_1.2,0.6695807933807373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,8,1,power_law_1.01,0.24391040802001954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,8,1,power_law_1.2,0.8497344017028808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,8,1,power_law_1.2,0.6745600223541259
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,8,1,power_law_1.01,0.8646207809448242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,8,1,power_law_1.01,0.27525761127471926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,8,1,power_law_1.2,1.0162495613098144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,8,1,power_law_1.2,0.7558911800384521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,8,1,power_law_1.01,0.3112191915512085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,8,1,power_law_1.01,1.1801088333129883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,balanced,0.45610666275024414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,8,1,power_law_1.2,1.1676544189453124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,balanced,0.5249919891357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,8,1,power_law_1.01,0.32163200378417967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,balanced,0.7611733277638754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,8,1,power_law_1.2,0.028332799673080444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,8,1,power_law_1.01,0.3333631992340088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,8,1,power_law_1.01,1.6693376541137694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,balanced,0.9060053030649821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,8,1,power_law_1.2,1.4773568153381347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,balanced,0.031248000760873158
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,8,1,power_law_1.2,0.03030399978160858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,balanced,0.031514666974544525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,balanced,0.03563733398914337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,8,1,power_law_1.01,0.36046719551086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,balanced,0.050197333097457886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,balanced,1.2870666980743408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,balanced,0.07520000139872234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,balanced,0.1149013340473175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,8,1,power_law_1.2,0.04009599983692169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,balanced,0.12708800037701926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,8,1,power_law_1.01,0.3731391906738281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,balanced,1.66157865524292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,8,1,power_law_1.2,0.060249602794647215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,8,1,power_law_1.2,0.05854079723358154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,8,1,power_law_1.01,0.40070400238037107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,balanced,0.04386133452256521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,balanced,0.04188799858093262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,8,1,power_law_1.2,0.9419136047363281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,8,1,power_law_1.2,0.07209600210189819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,balanced,0.05187733471393585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,8,1,power_law_1.2,0.07986559867858886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,balanced,0.06256533165772755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,balanced,0.07854400078455608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,balanced,0.12892799576123556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,balanced,0.12609600027402243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,8,1,power_law_1.2,0.1013375997543335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,balanced,0.13126933574676514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,8,1,power_law_1.2,0.11450879573822022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,8,1,power_law_1.2,0.9895744323730469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,balanced,0.21185066302617392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,balanced,0.21318932374318442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,balanced,0.21860800186793009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,8,1,power_law_1.2,0.152019202709198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,balanced,0.2216906746228536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,8,1,power_law_1.2,0.14421759843826293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,balanced,0.22702399889628092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,8,1,power_law_1.2,1.092454433441162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,balanced,0.23600000143051147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,balanced,0.24386133750279745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,balanced,0.12821333607037863
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,8,1,power_law_1.2,0.17129600048065186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,balanced,0.1274720033009847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,8,1,power_law_1.2,1.766796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,8,1,power_law_1.2,1.172659206390381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,8,1,power_law_1.2,0.17866239547729493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,8,1,power_law_1.01,0.4214208126068115
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,8,1,power_law_1.2,0.18309119939804078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,8,1,power_law_1.2,1.373465633392334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,8,1,power_law_1.01,0.4365248203277588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,8,1,power_law_1.2,2.3775871276855467
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,8,1,power_law_1.2,0.1937343955039978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,8,1,power_law_1.01,0.4598527908325195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,8,1,power_law_1.2,0.20202240943908692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,balanced,0.2527573307355245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,8,1,power_law_1.01,0.5250368118286133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,balanced,0.26929599046707153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,balanced,0.12987200419108072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,8,1,power_law_1.2,0.2130176067352295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,balanced,0.3009493350982666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,balanced,0.15758400162061056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,balanced,0.3306186596552531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,balanced,0.16342932979265848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,8,1,power_law_1.2,2.9812671661376955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,8,1,power_law_1.01,0.6304255962371826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,balanced,0.3906240065892537
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,8,1,power_law_1.2,0.2305216073989868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,balanced,0.4533173243204753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,balanced,0.6642293135325114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,8,1,power_law_1.2,0.24885120391845703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,balanced,0.7832533518473307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,8,1,balanced,0.7241919835408529
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,8,1,power_law_1.2,0.2663104057312012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,balanced,1.115221341451009
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,8,1,power_law_1.2,0.30189440250396726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,balanced,1.448560078938802
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,8,1,power_law_1.2,0.37932798862457273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,balanced,0.16686934232711792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,8,1,power_law_1.2,0.342739200592041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,balanced,2.1107254028320312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,8,1,power_law_1.2,0.4365568161010742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,8,1,power_law_1.2,0.4251584053039551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,balanced,2.7767680486043296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,8,1,power_law_1.2,0.2144063949584961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,8,1,power_law_1.2,0.5599167823791504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,8,1,power_law_1.2,0.2531775951385498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,8,1,power_law_1.2,0.6023808002471924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,8,1,power_law_1.2,0.30945279598236086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,balanced,0.17466133832931519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,balanced,0.1811413367589315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,power_law_1.01,0.059334397315979004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,balanced,0.18476800123850504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,8,1,power_law_1.2,0.7981760025024414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,balanced,0.19963733355204263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,8,1,power_law_1.2,0.31870079040527344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,balanced,0.2364906668663025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,power_law_1.01,0.09594240188598632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,balanced,0.25438400109608966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,balanced,0.295418659845988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,8,1,power_law_1.2,0.3309312105178833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,8,1,power_law_1.2,0.8927871704101562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,power_law_1.01,0.11198079586029053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,balanced,0.33530668417612713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,8,1,power_law_1.2,1.549516773223877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,power_law_1.01,0.1416640043258667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,balanced,0.4151093165079753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,8,1,power_law_1.2,0.3547647953033447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,balanced,0.49371198813120526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,8,1,power_law_1.2,1.2033535957336425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,balanced,0.7014719645182291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,balanced,0.02754133443037669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,balanced,0.02752533306678136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,balanced,0.029359998802344005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,balanced,0.8611199855804443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,balanced,0.03350399931271871
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,balanced,0.0499893327554067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,balanced,0.07014933228492737
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,balanced,0.07227733234564464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,balanced,1.225157340367635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,balanced,0.07483200232187907
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,8,1,power_law_1.2,1.7230783462524415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,balanced,0.07420266668001811
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,balanced,0.07403199871381123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,balanced,0.07521066566308339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,power_law_1.01,0.048204800486564635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,balanced,1.5893813769022624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,power_law_1.01,0.06716799736022949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,power_law_1.01,0.14640640020370482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,power_law_1.01,0.05857279896736145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,power_law_1.01,0.08638719916343689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,8,1,power_law_1.2,0.37637760639190676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,power_law_1.01,0.15247999429702758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,power_law_1.01,0.07297279834747314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,power_law_1.01,0.08399360179901123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,8,1,power_law_1.2,1.9239231109619142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,power_law_1.01,0.1611840009689331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,power_law_1.01,0.07828480005264282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,power_law_1.01,0.09623680114746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,8,1,power_law_1.2,0.40202879905700684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,power_law_1.01,0.157478404045105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,power_law_1.01,0.0953984022140503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,power_law_1.01,0.09754239916801452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,8,1,power_law_1.2,0.42026238441467284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,power_law_1.01,0.18120959997177125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,power_law_1.01,0.09715200066566468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,balanced,0.07499733567237854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,8,1,power_law_1.2,2.2832704544067384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,8,1,power_law_1.2,0.43877758979797366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,power_law_1.01,0.17828480005264283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,power_law_1.01,0.09867519736289979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,power_law_1.01,0.18495359420776367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,power_law_1.01,0.11146240234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,8,1,power_law_1.2,0.4579648017883301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,power_law_1.01,0.19585920572280885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,power_law_1.01,0.12471679449081421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,8,1,power_law_1.2,0.530457592010498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,power_law_1.01,0.20536320209503173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,power_law_1.01,0.1982208013534546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,8,1,power_law_1.2,3.0205888748168945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,power_law_1.01,0.20417919158935546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,power_law_1.01,0.19943679571151735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,8,1,power_law_1.2,0.6317887783050538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,power_law_1.01,0.21068799495697021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,power_law_1.01,0.19831680059432982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,power_law_1.01,0.11288959980010986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,power_law_1.01,0.22283520698547363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,power_law_1.01,0.20773758888244628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,balanced,0.07682666679223378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,power_law_1.01,0.1146496057510376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,power_law_1.01,0.24578559398651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,power_law_1.01,0.21659519672393798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,power_law_1.01,0.12242560386657715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,power_law_1.01,0.2847424030303955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,power_law_1.01,0.22603518962860109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,power_law_1.01,0.12337919473648071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,8,1,power_law_1.2,3.73741455078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,power_law_1.01,0.30950400829315183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,power_law_1.01,0.2327807903289795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,power_law_1.01,0.14381439685821534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,power_law_1.01,0.37448959350585936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,power_law_1.01,0.24180479049682618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,power_law_1.01,0.149344003200531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,power_law_1.01,0.2761280059814453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,power_law_1.01,0.4247231960296631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,power_law_1.01,0.15652480125427246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,power_law_1.01,0.3159679889678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,power_law_1.01,0.16312960386276246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,power_law_1.01,0.5261760234832764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,8,1,power_law_1.01,0.7754111766815186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,power_law_1.01,0.3547840118408203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,power_law_1.01,0.17010560035705566
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,balanced,0.07798400024573009
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,balanced,0.10492799679438274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,power_law_1.01,0.6308095932006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,balanced,0.10083199540774028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,power_law_1.01,0.18167680501937866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,power_law_1.01,0.4463168144226074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,8,1,power_law_1.01,0.945958423614502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,power_law_1.01,0.19296640157699585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,power_law_1.01,0.8623871803283691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,power_law_1.01,0.5470719814300538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,power_law_1.01,0.2316672086715698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,power_law_1.01,0.25622398853302003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,power_law_1.01,0.7605631828308106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,power_law_1.01,1.064249610900879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,power_law_1.01,0.3067008018493652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,power_law_1.01,0.926035213470459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,power_law_1.01,0.35766398906707764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,power_law_1.01,1.5118592262268067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,power_law_1.01,0.029798400402069092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,power_law_1.01,0.45725440979003906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,balanced,0.09044800202051799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,power_law_1.01,1.3221311569213867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,power_law_1.01,0.046540799736976626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,balanced,0.17292267084121704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,balanced,0.1188320020834605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,power_law_1.01,0.5474751949310303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,power_law_1.01,0.052236801385879515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,power_law_1.01,1.9799104690551759
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,power_law_1.01,0.05413119792938233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,power_law_1.01,1.6752832412719727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,power_law_1.2,0.05948160290718078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,power_law_1.01,0.062483197450637816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,power_law_1.2,0.09559040069580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,power_law_1.01,0.05716480016708374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,power_law_1.2,0.09400320053100586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,power_law_1.01,0.06410880088806152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,power_law_1.01,2.392019271850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,power_law_1.2,0.12340480089187622
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,power_law_1.01,0.06696959733963012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,power_law_1.2,0.14337919950485228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,balanced,0.11116799712181091
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,balanced,0.1199679970741272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,power_law_1.2,0.12300159931182861
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,balanced,0.14013333121935526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,balanced,0.1620373328526815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,power_law_1.01,0.7393087863922119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,power_law_1.2,0.14648319482803346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,balanced,0.22735466559727988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,balanced,0.2739413380622864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,power_law_1.01,3.2468544006347657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,balanced,0.3879306713740031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,power_law_1.2,0.15877120494842528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,power_law_1.01,0.9481856346130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,balanced,0.5015413363774618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,power_law_1.2,0.17802879810333253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,8,1,balanced,1.0497013727823894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,power_law_1.2,0.18526079654693603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,power_law_1.2,0.05921919941902161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,power_law_1.01,1.3546943664550781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,power_law_1.2,0.19417599439620972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,power_law_1.2,0.07164160013198853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,power_law_1.01,0.0662335991859436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,power_law_1.2,0.18984320163726806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,power_law_1.2,0.07288960218429566
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,power_law_1.01,0.06728960275650024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,power_law_1.2,0.1944383978843689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,power_law_1.2,0.08235520124435425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,power_law_1.01,1.7152959823608398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,power_law_1.01,0.07025920152664185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,power_law_1.2,0.11238399744033814
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,power_law_1.01,0.06833919882774353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,power_law_1.2,0.04880639910697937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,power_law_1.2,0.10210560560226441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,power_law_1.01,0.06981120109558106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,power_law_1.2,0.06837120056152343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,power_law_1.01,0.07301759719848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,power_law_1.2,0.11811200380325318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,power_law_1.2,0.06272000074386597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,power_law_1.01,0.0821183979511261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,power_law_1.2,0.11688319444656373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,power_law_1.2,0.08244479894638061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,power_law_1.01,0.08241279721260071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,power_law_1.2,0.19420160055160524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,power_law_1.2,0.09488639831542969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,power_law_1.01,0.08382080197334289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,power_law_1.2,0.19468799829483033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,power_law_1.01,0.10865919589996338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,power_law_1.2,0.09807999730110169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,power_law_1.2,0.19752960205078124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,power_law_1.2,0.11585919857025147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,power_law_1.2,0.2093183994293213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,power_law_1.2,0.11152640581130982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,power_law_1.2,0.20767359733581542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,power_law_1.2,0.10818560123443603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,power_law_1.2,0.22408320903778076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,power_law_1.2,0.11755520105361938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,power_law_1.2,0.23443200588226318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,power_law_1.2,0.028883200883865357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,power_law_1.2,0.14614399671554565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,power_law_1.2,0.2506432056427002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,8,1,power_law_1.2,0.7802559852600097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,power_law_1.2,0.0442111998796463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,power_law_1.2,0.15364480018615723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,power_law_1.2,0.27949440479278564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,power_law_1.2,0.20421760082244872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,power_law_1.2,0.15569920539855958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,power_law_1.01,0.10257920026779174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,power_law_1.2,0.3298304080963135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,power_law_1.2,0.2206399917602539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,power_law_1.2,0.16620160341262818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,power_law_1.01,0.11316479444503784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,power_law_1.2,0.2350912094116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,power_law_1.2,0.3731328010559082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,power_law_1.01,0.12918399572372435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,power_law_1.2,0.2578687906265259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,power_law_1.2,0.45537919998168946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,power_law_1.01,0.15621119737625122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,power_law_1.2,0.3034559965133667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,power_law_1.2,0.5811264038085937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,power_law_1.01,0.18700159788131715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,power_law_1.2,0.33349759578704835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,power_law_1.01,0.25560319423675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,power_law_1.2,0.8003840446472168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,power_law_1.2,0.3866944074630737
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,power_law_1.2,0.047040000557899475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,power_law_1.01,0.32952959537506105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,power_law_1.2,0.44077439308166505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,power_law_1.2,0.053395199775695804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,power_law_1.2,0.9873279571533203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,power_law_1.01,0.46034560203552244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,power_law_1.2,0.1712448000907898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,power_law_1.2,0.5381696224212646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,power_law_1.2,0.06175360083580017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,power_law_1.2,0.17967360019683837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,power_law_1.01,0.6935935974121094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,power_law_1.2,1.3541312217712402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,power_law_1.2,0.2023168087005615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,power_law_1.2,0.057004797458648684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,power_law_1.2,0.6432191848754882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,balanced,0.025424001117547352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,balanced,0.029306667546431225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,balanced,0.0436160018046697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,balanced,0.06948266426722209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,power_law_1.2,0.23601920604705812
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,power_law_1.2,0.06616960167884826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,balanced,0.12545067071914673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,balanced,0.22738667329152426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,balanced,0.22614399592081705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,power_law_1.2,0.9095744132995606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,8,1,balanced,1.1778346697489421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,balanced,0.2262986699740092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,power_law_1.2,1.6889024734497071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,8,1,balanced,1.8037707010904949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,power_law_1.2,1.0980735778808595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,8,1,balanced,2.0308639208475747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,power_law_1.2,2.5185855865478515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,power_law_1.2,1.536025619506836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,8,1,balanced,2.791797320048014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,power_law_1.2,0.2607167959213257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,power_law_1.2,0.06414719820022582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,balanced,0.20923733711242676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,power_law_1.2,2.065779113769531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,power_law_1.2,0.31719679832458497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,balanced,0.21025067567825317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,power_law_1.2,0.06918399930000305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,balanced,0.025274666647116344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,balanced,0.2069759964942932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,balanced,0.02738133321205775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,balanced,0.20595200856526694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,balanced,0.03332266708215078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,8,1,power_law_1.2,0.9868032455444335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,balanced,0.04614399870236715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,power_law_1.2,0.3683775901794434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,balanced,0.2046133279800415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,power_law_1.2,0.06962559819221496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,balanced,0.08080000181992848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,balanced,0.20965866247812906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,balanced,0.1325920025507609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,balanced,0.21278399229049683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,balanced,0.15574933091799417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,balanced,0.2151040037473043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,8,1,balanced,3.8010613123575845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,power_law_1.2,0.4739327907562256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,balanced,0.22081599632898966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,balanced,0.28697067499160767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,balanced,0.29679999748865765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,power_law_1.2,0.5638720035552979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,balanced,0.318938672542572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,balanced,0.33818666140238446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,balanced,0.38171199957529706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,power_law_1.2,0.783846378326416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,balanced,0.4222613175710042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,8,1,power_law_1.01,1.1892352104187012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,8,1,balanced,5.340842564900716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,power_law_1.2,0.9787455558776855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,power_law_1.2,3.4138431549072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,8,1,power_law_1.01,1.4839296340942383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,power_law_1.01,0.02426239997148514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,balanced,0.15506666898727417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,power_law_1.2,0.06692479848861695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,balanced,0.14827199776967367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,balanced,0.1495519975821177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,balanced,0.15616533160209656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,power_law_1.2,0.06837760210037232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,balanced,0.15620799859364828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,balanced,0.1581013302008311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,balanced,0.1606613298257192
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,power_law_1.2,0.06895359754562377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,8,1,power_law_1.01,1.775257682800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,balanced,0.16403200229008993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,balanced,0.16874667008717856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,power_law_1.2,0.07178879976272583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,balanced,0.1745013395945231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,balanced,0.23321600755055746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,balanced,0.24338134129842123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,power_law_1.2,0.07714560031890869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,balanced,0.6492000023523966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,balanced,0.26552534103393555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,balanced,0.2908266584078471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,balanced,0.7389813264211019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,balanced,0.3372693459192912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,8,1,power_law_1.01,2.3336896896362305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,balanced,0.38233065605163574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,balanced,1.0509920120239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,balanced,0.581658681233724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,power_law_1.01,0.02905600070953369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,balanced,1.3704160054524739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,balanced,0.6759946346282959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,balanced,0.025306666890780132
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,balanced,0.02720000098148982
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,balanced,0.028064000109831493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,power_law_1.01,0.02361599951982498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,balanced,0.03331733246644338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,balanced,0.959706703821818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,balanced,0.05217599868774414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,balanced,0.0786186655362447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,balanced,0.07824533184369405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,8,1,power_law_1.01,2.9332799911499023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,balanced,0.08066666622956593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,balanced,1.2537813186645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,balanced,0.0819893330335617
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,balanced,0.08433600266774495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,balanced,0.0867680013179779
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,balanced,0.08943999807039897
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,balanced,0.09288533528645833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,power_law_1.2,0.08009600043296813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,balanced,0.09503466884295146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,balanced,0.1074186662832896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,balanced,0.11170132954915364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,power_law_1.2,0.08243839740753174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,balanced,0.1227839986483256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,balanced,0.12609066565831503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,balanced,0.13780267039934793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,power_law_1.2,0.1223296046257019
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,balanced,0.14755200346310934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,power_law_1.2,1.392140769958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,balanced,0.1666719913482666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,balanced,0.22842667500178018
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,power_law_1.2,0.1070207953453064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,power_law_1.01,0.044095999002456664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,balanced,0.24925333261489868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,balanced,0.41917868455251056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,power_law_1.2,0.11715840101242066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,power_law_1.01,0.0681984007358551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,power_law_1.01,0.025235199928283693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,balanced,0.36899733543395996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,power_law_1.2,1.741708755493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,balanced,0.520309329032898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,power_law_1.01,0.1060479998588562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,power_law_1.01,0.033855998516082765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,balanced,0.6714186668395996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,power_law_1.01,0.13146239519119263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,power_law_1.01,0.04694400131702423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,power_law_1.2,0.02282879948616028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,power_law_1.01,0.16510080099105834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,power_law_1.01,0.06618239879608154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,power_law_1.2,0.024915200471878052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,power_law_1.01,0.1741312026977539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,power_law_1.01,0.08650879859924317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,power_law_1.2,0.034694400429725644
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,power_law_1.01,0.02542079985141754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,power_law_1.01,0.16945279836654664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,8,1,power_law_1.01,4.040985488891602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,power_law_1.01,0.12297600507736206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,power_law_1.2,0.04556800127029419
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,power_law_1.01,0.025273600220680238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,power_law_1.01,0.17380479574203492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,power_law_1.01,0.1282495975494385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,power_law_1.2,0.06273919939994813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,power_law_1.01,0.027820798754692077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,power_law_1.01,0.12050559520721435
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,power_law_1.2,0.133241605758667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,power_law_1.2,0.07824640274047852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,power_law_1.01,0.036025598645210266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,power_law_1.01,0.12412159442901612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,power_law_1.2,0.023871999979019166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,power_law_1.2,0.1214400053024292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,power_law_1.2,0.1775231957435608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,8,1,power_law_1.2,1.2438912391662598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,power_law_1.01,0.044819200038909913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,power_law_1.2,0.02849920094013214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,power_law_1.2,0.1257024049758911
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,power_law_1.2,0.2008768081665039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,power_law_1.2,0.04472320079803467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,power_law_1.2,0.12027519941329956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,power_law_1.2,0.2678656101226807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,power_law_1.2,0.06568959951400757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,power_law_1.2,0.12431999444961547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,8,1,power_law_1.01,5.158732986450195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,power_law_1.2,0.34028799533843995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,power_law_1.2,0.1416767954826355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,power_law_1.2,0.14250880479812622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,power_law_1.01,0.18510719537734985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,power_law_1.2,0.4878079891204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,power_law_1.2,0.14488320350646972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,power_law_1.01,0.18558080196380616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,power_law_1.2,0.15222400426864624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,power_law_1.2,0.7671040058135986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,power_law_1.01,0.18721920251846313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,power_law_1.2,0.1555840015411377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,power_law_1.01,0.14010239839553834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,power_law_1.01,0.1920639991760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,power_law_1.2,0.16132479906082153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,power_law_1.01,0.14382719993591309
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,power_law_1.01,0.06110079884529114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,power_law_1.01,0.19658880233764647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,power_law_1.2,0.17065600156784058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,power_law_1.01,0.14382079839706421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,power_law_1.01,0.20181119441986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,power_law_1.2,0.23342080116271974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,power_law_1.01,0.14967679977416992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,power_law_1.2,0.0986303985118866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,power_law_1.2,0.24821119308471679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,power_law_1.01,0.15457919836044312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,power_law_1.2,0.12124799489974976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,power_law_1.2,0.2921087980270386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,power_law_1.01,0.15863679647445678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,power_law_1.2,0.16319359540939332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,power_law_1.01,0.1681023955345154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,power_law_1.2,0.32433280944824217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,power_law_1.2,0.023391999304294586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,power_law_1.2,0.16706559658050538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,power_law_1.01,0.22608640193939208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,power_law_1.2,0.39188480377197266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,power_law_1.2,0.024512000381946564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,power_law_1.2,0.1584895968437195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,power_law_1.01,0.24067199230194092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,power_law_1.2,0.461568021774292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,power_law_1.2,0.02710399925708771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,power_law_1.2,0.16752640008926392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,power_law_1.01,0.27409920692443845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,power_law_1.01,0.06997759938240052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,power_law_1.2,0.03318400084972382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,power_law_1.2,0.6122303962707519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,power_law_1.2,0.18524160385131835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,power_law_1.01,0.3180288076400757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,power_law_1.01,0.20904319286346434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,power_law_1.01,0.07379840016365051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,power_law_1.2,0.04195199906826019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,power_law_1.2,0.18568960428237916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,power_law_1.01,0.37747199535369874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,power_law_1.01,0.27596800327301024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,power_law_1.01,0.07597439885139465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,power_law_1.2,0.1871616005897522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,power_law_1.2,0.058233600854873654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,power_law_1.01,0.29297919273376466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,power_law_1.01,0.45593600273132323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,power_law_1.01,0.0767359972000122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,power_law_1.2,0.19330559968948363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,power_law_1.2,0.06732800006866455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,power_law_1.01,0.32522881031036377
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,power_law_1.01,0.08112639784812928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,power_law_1.01,0.6026944160461426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,power_law_1.2,0.07052800059318542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,power_law_1.01,0.37852799892425537
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,power_law_1.01,0.08384000062942505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,power_law_1.2,0.07291520237922669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,power_law_1.01,0.7378111839294433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,power_law_1.01,0.4375487804412842
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,power_law_1.01,0.08860160112380981
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,power_law_1.2,0.07588480114936828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,power_law_1.01,0.516377592086792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,power_law_1.01,0.09405440092086792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,power_law_1.2,0.08022400140762329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,power_law_1.01,1.0337023735046387
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,power_law_1.01,0.10081919431686401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,power_law_1.01,0.6784512042999268
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,power_law_1.2,0.08333439826965332
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,power_law_1.01,0.10744960308074951
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,power_law_1.2,0.08648319840431214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,power_law_1.01,1.325049591064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,power_law_1.01,0.8267392158508301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,power_law_1.2,0.09292799830436707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,power_law_1.2,0.19891200065612794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,power_law_1.2,0.10077439546585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,power_law_1.2,0.20698239803314208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,power_law_1.01,1.1383423805236816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,power_law_1.2,0.10776319503784179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,power_law_1.2,0.21631360054016113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,power_law_1.2,0.12510080337524415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,power_law_1.2,0.2832895994186401
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,power_law_1.2,0.1219264030456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,power_law_1.01,1.4591744422912598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,power_law_1.2,0.30278398990631106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,power_law_1.2,0.7542143821716308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,power_law_1.2,0.34919679164886475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,power_law_1.01,0.12420480251312256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,power_law_1.2,1.0491904258728026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,8,1,power_law_1.2,1.5418815612792969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,power_law_1.01,0.12121599912643433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,power_law_1.01,0.1397696018218994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,power_law_1.2,1.3450943946838378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,power_law_1.01,0.14525439739227294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,8,1,power_law_1.2,1.8089408874511719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,power_law_1.01,0.17256319522857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,power_law_1.01,0.21969919204711913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,power_law_1.2,0.1401535987854004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,power_law_1.2,0.14728959798812866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,power_law_1.01,0.2682496070861816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,8,1,power_law_1.2,2.399737548828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,power_law_1.2,0.3855936050415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,power_law_1.01,0.31994879245758057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,power_law_1.2,0.45168638229370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,power_law_1.01,0.3982399940490723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,power_law_1.2,0.5277440071105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,power_law_1.01,0.547596788406372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,8,1,power_law_1.2,2.94451847076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,power_law_1.2,0.6842175960540772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,power_law_1.01,0.7684095859527588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,power_law_1.2,0.840396785736084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,power_law_1.2,0.17375359535217286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,power_law_1.2,0.2218048095703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,power_law_1.2,0.2720128059387207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,8,1,power_law_1.2,4.078630447387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,power_law_1.2,0.3253247976303101
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,power_law_1.2,0.4041920185089111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,power_law_1.2,0.5521024227142334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,8,balanced,0.033610666791598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,8,balanced,0.03751999884843826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,8,balanced,0.0412266676624616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,8,balanced,0.05995733539263407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,power_law_1.2,1.1478912353515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,power_law_1.2,0.7844223976135254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,8,1,power_law_1.2,5.198271942138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,8,power_law_1.01,0.04031359851360321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,power_law_1.2,1.4735936164855956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,8,power_law_1.01,0.0578495979309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,8,power_law_1.01,0.05691519975662231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,8,power_law_1.01,0.06123520135879516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,8,power_law_1.01,0.06876159906387329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,8,power_law_1.01,0.06709120273590088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,8,balanced,0.08893866340319316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,8,balanced,0.11176000038782756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,8,power_law_1.01,0.08381440043449402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,8,balanced,0.1239466667175293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,8,balanced,0.1232373317082723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,8,balanced,0.11385066310564677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,8,balanced,0.11488533020019531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,8,balanced,0.12785599629084268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,8,balanced,0.12986666957537332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,8,balanced,0.13051733374595642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,8,balanced,0.1377173364162445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,8,balanced,0.1425493359565735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,8,balanced,0.14662399888038635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,8,balanced,0.15617600083351135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,8,balanced,0.18521066506703696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,8,balanced,0.20213866233825684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,8,balanced,0.23271999756495157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,8,balanced,0.2633226712544759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,8,balanced,0.37546666463216144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,8,balanced,0.4352533419926961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,8,balanced,0.03166933357715607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,8,balanced,0.033733333150545754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,8,balanced,0.03569599986076355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,8,balanced,0.050026665131251015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,8,balanced,0.06469333171844482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,8,power_law_1.01,0.08457599878311158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,8,balanced,0.07840533554553986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,8,balanced,0.09106666843096416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,8,power_law_1.01,0.08741120100021363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,8,power_law_1.01,0.052767997980117796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,8,power_law_1.01,0.08743680119514466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,8,power_law_1.2,0.03899520039558411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,8,power_law_1.01,0.08097919821739197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,8,power_law_1.01,0.09514880180358887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,8,power_law_1.2,0.053324800729751584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,8,power_law_1.01,0.10067839622497558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,8,power_law_1.01,0.077811199426651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,8,power_law_1.2,0.052275198698043826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,8,power_law_1.01,0.10168960094451904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,8,power_law_1.01,0.09127680063247681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,8,power_law_1.2,0.06033279895782471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,8,power_law_1.01,0.11059199571609497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,8,power_law_1.01,0.09562240242958069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,8,power_law_1.2,0.07150080204010009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,8,power_law_1.01,0.11460479497909545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,8,power_law_1.01,0.09632639884948731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,8,balanced,0.6033066511154175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,8,power_law_1.2,0.07169920206069946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,8,power_law_1.01,0.12124160528182984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,8,power_law_1.01,0.10846079587936401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,8,power_law_1.2,0.07982720136642456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,8,balanced,0.7260053157806396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,8,power_law_1.01,0.13270399570465088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,8,power_law_1.01,0.11511679887771606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,8,power_law_1.2,0.08163840174674988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,8,balanced,0.0929813285668691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,8,power_law_1.01,0.15860480070114136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,8,power_law_1.01,0.10925439596176148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,8,power_law_1.2,0.08627200126647949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,8,power_law_1.01,0.1826367974281311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,8,power_law_1.01,0.10696320533752442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,8,power_law_1.2,0.08428800106048584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,8,power_law_1.2,0.05374079942703247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,8,power_law_1.01,0.11429760456085206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,8,power_law_1.2,0.06828160285949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,8,power_law_1.01,0.1186560034751892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,8,power_law_1.2,0.0753600001335144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,8,power_law_1.01,0.11756160259246826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,8,power_law_1.2,0.08081920146942138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,8,power_law_1.01,0.13256959915161132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,8,power_law_1.2,0.09056640267372132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,8,power_law_1.01,0.14083839654922486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,8,power_law_1.2,0.09659519791603088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,8,power_law_1.01,0.14206720590591432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,8,power_law_1.2,0.10980479717254639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,8,power_law_1.01,0.1611456036567688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,8,balanced,0.08804800113042195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,8,power_law_1.2,0.10938880443572999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,8,balanced,0.08940266569455464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,8,power_law_1.01,0.18986239433288574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,8,balanced,0.10048000017801921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,8,balanced,0.10545066992441814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,8,power_law_1.2,0.10951679944992065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,8,power_law_1.01,0.21715199947357178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,8,balanced,0.10707733035087585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,8,balanced,0.11319466431935628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,8,power_law_1.2,0.10781439542770385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,8,balanced,0.11931733290354411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,8,balanced,0.12505599856376648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,8,power_law_1.01,0.2566783905029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,8,balanced,0.13391466935475668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,8,power_law_1.2,0.11964160203933716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,8,balanced,0.15680000185966492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,8,power_law_1.01,0.31123199462890627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,8,balanced,0.17364267508188883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,8,power_law_1.2,0.12403839826583862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,8,balanced,0.21134400367736816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,8,balanced,0.24493332703908285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,8,power_law_1.2,0.12613120079040527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,8,power_law_1.2,0.13557759523391724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,8,1,balanced,0.03153600047032038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,8,power_law_1.2,0.14211200475692748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,8,1,balanced,0.045466666420300804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,8,1,balanced,0.06555200119813283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,8,power_law_1.2,0.14710400104522706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,8,balanced,1.0619200070699055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,8,power_law_1.2,0.16216319799423218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,8,power_law_1.2,0.1956928014755249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,8,balanced,1.3555466334025066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,8,power_law_1.2,0.09697920083999634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,8,power_law_1.2,0.22116479873657227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,8,power_law_1.2,0.09927679896354676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,8,1,balanced,0.031471999982992806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,8,1,balanced,0.04178666571776072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,8,power_law_1.2,0.2671040058135986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,8,1,balanced,0.06418133278687795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,8,power_law_1.2,0.10375679731369018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,8,power_law_1.2,0.32936320304870603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,8,power_law_1.2,0.10712319612503052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,8,power_law_1.2,0.11724159717559815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,8,power_law_1.2,0.12259839773178101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,8,power_law_1.2,0.1367616057395935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,8,power_law_1.2,0.16257920265197753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,8,power_law_1.2,0.18611840009689332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,8,power_law_1.01,0.22676479816436768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,8,power_law_1.2,0.23145599365234376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,8,power_law_1.01,0.2673664093017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,8,power_law_1.2,0.2788160085678101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,8,power_law_1.01,0.36116480827331543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,8,power_law_1.01,0.440934419631958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,8,power_law_1.01,0.6158847808837891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,8,power_law_1.01,0.7694975852966308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,8,power_law_1.01,0.38787839412689207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,8,power_law_1.01,0.4878975868225098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,8,power_law_1.01,1.1124608039855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,8,balanced,0.3434133529663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,8,power_law_1.01,0.6855103969573975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,8,balanced,0.4110666513442993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,8,balanced,0.5725013415018717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,8,power_law_1.01,1.477670383453369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,8,power_law_1.01,0.8752063751220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,8,balanced,0.7114026546478271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,8,balanced,1.034816026687622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,8,power_law_1.01,1.2503423690795898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,8,balanced,1.3340959548950195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,8,power_law_1.2,0.41707520484924315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,8,power_law_1.01,1.6933439254760743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,8,power_law_1.2,0.5270592212677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,8,1,balanced,0.11948800086975098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,8,1,balanced,0.21372799078623453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,8,1,balanced,0.2826240062713623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,8,power_law_1.2,0.754201602935791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,8,1,balanced,0.1095360020796458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,8,1,balanced,0.19779733816782633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,8,1,balanced,0.25778132677078247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,8,power_law_1.2,0.37077760696411133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,8,power_law_1.2,0.45048961639404295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,8,power_law_1.2,0.6611392021179199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,8,power_law_1.2,0.8236160278320312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,8,power_law_1.2,1.173344039916992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,8,power_law_1.2,1.5019776344299316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,8,balanced,0.06428266565004985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,8,balanced,0.11796266833941142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,8,balanced,0.16267200311024985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,8,balanced,0.30170132716496784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,8,1,power_law_1.01,0.030022400617599487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,8,balanced,0.5234080155690511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,8,1,power_law_1.01,0.04188160002231598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,8,1,power_law_1.01,0.06771199703216553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,8,1,power_law_1.2,0.02969599962234497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,8,1,power_law_1.2,0.04631040096282959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,8,1,power_law_1.2,0.06759039759635925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,8,balanced,0.6857759952545166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,8,power_law_1.2,0.9473024368286133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,8,balanced,0.7020213603973389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,8,balanced,0.7094026406606039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,8,balanced,0.7378506660461426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,8,power_law_1.2,1.3638527870178223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,8,balanced,0.7429013252258301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,8,balanced,0.8610080083211263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,8,balanced,0.880511999130249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,8,1,balanced,0.30770132939020794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,8,balanced,0.8977973461151123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,8,1,balanced,0.3081013361612956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,8,1,power_law_1.01,0.030451199412345885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,8,balanced,0.9194080034891764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,8,1,power_law_1.01,0.04142079949378967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,8,1,power_law_1.01,0.06670719981193543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,8,balanced,0.9419253667195638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,8,balanced,0.9661920070648193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,8,balanced,1.0111839771270752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,8,1,power_law_1.2,0.030297601222991945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,8,balanced,1.1638826529184978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,8,1,power_law_1.2,0.040838399529457094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,8,1,balanced,0.302021324634552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,8,1,balanced,0.3014773329099019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,8,balanced,1.2400800387064617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,8,balanced,1.3876159985860188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,8,power_law_1.2,1.8383295059204101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,8,1,power_law_1.2,0.0660863995552063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,8,1,power_law_1.01,0.10417920351028442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,8,1,power_law_1.01,0.14261120557785034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,8,balanced,1.5376052856445312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,8,1,power_law_1.01,0.22001919746398926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,8,balanced,2.2569707234700522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,8,1,power_law_1.2,0.09969919919967651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,8,1,power_law_1.2,0.12710399627685548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,8,balanced,2.5630772908528647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,8,1,power_law_1.2,0.21471359729766845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,8,balanced,0.05499200026194254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,8,balanced,0.0749066670735677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,8,balanced,0.12129599849383037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,8,balanced,0.18997333447138467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,8,balanced,3.6115201314290366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,8,balanced,0.2967680096626282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,8,balanced,0.39022934436798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,8,balanced,0.44548265139261883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,8,balanced,0.449402650197347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,8,balanced,0.422266681989034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,8,balanced,0.4291359980901082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,8,1,power_law_1.01,0.09724159836769104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,8,balanced,0.5666293303171793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,8,balanced,4.310538609822591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,8,balanced,0.5864106814066569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,8,1,power_law_1.01,0.13423999547958373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,8,balanced,0.6029173135757446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,8,1,power_law_1.01,0.20056960582733155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,8,balanced,0.628496011098226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,8,balanced,0.6517333189646403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,8,balanced,0.6770186424255371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,8,balanced,0.7299520174662272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,8,balanced,0.8403147061665853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,8,balanced,0.9245279630025228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,8,balanced,1.0879093011220295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,8,1,balanced,0.4238080183664958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,8,balanced,1.2503626346588135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,8,1,balanced,0.42402132352193195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,8,balanced,1.7679626146952312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,8,1,power_law_1.2,0.09336959719657897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,8,1,power_law_1.2,0.1245695948600769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,8,1,power_law_1.2,0.19567999839782715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,8,1,power_law_1.01,0.25639679431915285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,8,balanced,2.091525395711263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,8,1,power_law_1.01,0.2721280097961426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,8,balanced,2.954165458679199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,8,1,power_law_1.2,0.25692799091339114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,8,balanced,6.4650452931722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,8,1,power_law_1.2,0.2678976058959961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,8,balanced,3.6153599421183267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,8,balanced,8.214975992838541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,8,1,power_law_1.01,0.24232959747314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,8,1,power_law_1.01,0.24824960231781007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,8,1,power_law_1.2,0.24326400756835936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,8,1,balanced,1.268671989440918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,8,1,power_law_1.2,0.2566976070404053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,8,1,balanced,1.267797311147054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,8,1,power_law_1.01,0.3569983959197998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,8,1,power_law_1.01,0.36424961090087893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,8,balanced,5.463861465454102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,8,balanced,0.07400533556938171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,8,balanced,0.08416000008583069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,8,balanced,0.1091306706269582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,8,1,power_law_1.2,0.35148799419403076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,8,1,power_law_1.2,0.36934399604797363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,8,balanced,7.031359990437825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,8,power_law_1.01,0.2551935911178589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,8,power_law_1.01,0.45582079887390137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,8,power_law_1.01,0.39676799774169924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,8,1,balanced,0.41165868441263836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,8,power_law_1.01,0.5270463943481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,8,power_law_1.01,0.5749567985534668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,8,power_law_1.01,0.5921855926513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,8,balanced,0.19206400712331137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,8,power_law_1.01,0.6121920108795166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,8,balanced,0.30949334303538006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,8,power_law_1.01,0.60730881690979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,8,power_law_1.01,0.6109248161315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,8,power_law_1.01,0.6875199794769287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,8,power_law_1.01,0.7921216011047363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,8,power_law_1.01,0.8142335891723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,8,power_law_1.01,0.8430784225463868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,8,power_law_1.01,0.8703295707702636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,8,power_law_1.01,0.8959936141967774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,8,power_law_1.01,0.946291160583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,8,power_law_1.01,1.0047103881835937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,8,balanced,0.39715198675791424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,8,power_law_1.01,1.1589247703552246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,8,power_law_1.01,1.3206656455993653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,8,1,balanced,0.32201600074768066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,8,1,balanced,0.3216266632080078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,8,1,balanced,0.325546662012736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,8,1,balanced,0.3274400035540263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,8,1,balanced,0.33032000064849854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,8,1,balanced,0.3349173466364543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,8,balanced,0.40591998895009357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,8,balanced,0.412773331006368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,8,1,power_law_1.01,1.112441635131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,8,power_law_1.01,1.5298303604125976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,8,1,power_law_1.01,1.1255552291870117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,8,power_law_1.01,1.8175039291381836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,8,power_law_1.01,2.2379583358764648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,8,1,power_law_1.01,0.37896320819854734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,8,1,power_law_1.2,1.1302335739135743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,8,power_law_1.01,2.7976255416870117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,8,1,power_law_1.2,1.1713279724121093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,8,1,power_law_1.2,0.3814079999923706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,8,power_law_1.01,3.8523521423339844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,8,power_law_1.01,4.8897663116455075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,8,1,balanced,0.35256532828013104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,8,1,balanced,0.3614826599756877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,8,balanced,0.7813066641489664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,8,1,balanced,0.3842613299687703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,8,1,balanced,0.4073653221130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,8,balanced,0.7921706835428873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,8,1,balanced,0.6908640066782633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,8,balanced,0.8099733193715414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,8,1,balanced,0.7279520034790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,8,balanced,0.8304373423258463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,8,1,balanced,1.0452586809794109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,8,balanced,0.852351983388265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,8,1,balanced,1.14683198928833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,8,balanced,0.8836959997812907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,8,balanced,0.9164746602376302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,8,1,balanced,1.7781492869059246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,8,balanced,0.952735980351766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,8,balanced,1.0215413570404053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,8,1,balanced,2.1972692807515464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,8,balanced,1.1460533142089844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,8,1,power_law_1.01,0.2913343906402588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,8,balanced,1.2753492991129558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,8,1,power_law_1.01,0.286028790473938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,8,1,power_law_1.01,0.31383678913116453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,8,balanced,1.9453493754069011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,8,1,power_law_1.01,0.31712639331817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,8,1,power_law_1.01,0.33355519771575926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,8,1,power_law_1.2,0.2974720001220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,8,balanced,2.1985546747843423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,8,1,power_law_1.01,0.3508095979690552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,8,1,power_law_1.2,0.30681600570678713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,8,1,power_law_1.2,0.3169663906097412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,8,1,power_law_1.2,0.3285248041152954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,8,balanced,3.1416587829589844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,8,1,power_law_1.2,0.3386431932449341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,8,1,power_law_1.2,0.3588032007217407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,8,balanced,3.6427733103434243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,8,power_law_1.01,6.758207702636719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,8,power_law_1.01,0.15319679975509642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,8,power_law_1.01,0.2604543924331665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,8,power_law_1.01,0.24117119312286378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,8,balanced,5.512378692626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,8,balanced,6.944474538167317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,8,power_law_1.01,8.975494384765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,8,power_law_1.01,0.2981951951980591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,8,power_law_1.01,0.3544640064239502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,8,balanced,10.282112121582031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,8,power_law_1.01,0.37309439182281495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,8,1,power_law_1.01,0.3575808048248291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,8,1,power_law_1.01,0.40019841194152833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,8,1,power_law_1.01,0.46414718627929685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,8,1,balanced,1.2264959812164307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,8,1,power_law_1.01,0.523795223236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,8,balanced,13.150992075602213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,8,1,power_law_1.2,0.3740031957626343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,8,1,power_law_1.01,0.6532864093780517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,8,power_law_1.01,0.16842880249023437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,8,power_law_1.2,0.25547521114349364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,8,1,power_law_1.01,0.7745855808258056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,8,power_law_1.01,0.282694411277771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,8,power_law_1.2,0.3339008092880249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,8,power_law_1.01,0.40762882232666015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,8,1,power_law_1.01,1.0433664321899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,8,power_law_1.01,0.423737621307373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,8,power_law_1.2,0.4194943904876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,8,1,power_law_1.01,1.2936448097229003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,8,power_law_1.2,0.48688640594482424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,8,1,power_law_1.2,0.41777920722961426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,8,1,power_law_1.2,0.4782911777496338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,8,1,power_law_1.01,1.8237312316894532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,8,power_law_1.2,0.6136640071868896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,8,1,power_law_1.2,0.5339839935302735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,8,1,power_law_1.2,0.6656383991241455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,8,1,power_law_1.01,2.3488256454467775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,8,1,power_law_1.2,0.7845888137817383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,8,power_law_1.2,0.15394560098648072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,8,1,power_law_1.2,1.0556672096252442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,8,power_law_1.2,0.19198720455169677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,8,power_law_1.2,0.22792320251464843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,8,power_law_1.2,0.6410304069519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,8,power_law_1.2,0.275603199005127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,8,power_law_1.01,0.24140799045562744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,8,power_law_1.2,0.6291007995605469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,8,power_law_1.2,0.32001280784606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,8,power_law_1.01,0.3019392013549805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,8,power_law_1.2,0.6098624229431152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,8,power_law_1.2,0.3520576000213623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,8,power_law_1.01,0.3420799970626831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,8,power_law_1.2,0.6332032203674316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,8,power_law_1.2,0.38075520992279055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,8,power_law_1.01,0.36535038948059084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,8,power_law_1.01,0.37405440807342527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,8,power_law_1.2,0.666860818862915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,8,power_law_1.2,0.416812801361084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,8,power_law_1.01,0.35941119194030763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,8,power_law_1.2,0.3659264087677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,8,power_law_1.2,0.8032959938049317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,8,power_law_1.01,0.39830400943756106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,8,power_law_1.2,0.39013760089874266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,8,power_law_1.2,0.8393024444580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,8,power_law_1.01,0.7241600036621094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,8,power_law_1.2,0.5375679969787598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,8,power_law_1.2,0.8492416381835938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,8,power_law_1.01,0.7306047916412354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,8,power_law_1.2,0.5582208156585693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,8,power_law_1.2,0.8972352027893067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,8,power_law_1.2,0.5795519828796387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,8,power_law_1.01,0.757260799407959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,8,power_law_1.01,0.3898623943328857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,8,power_law_1.2,0.9155839920043946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,8,power_law_1.2,0.6119296073913574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,8,power_law_1.01,0.7904831886291503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,8,power_law_1.01,0.8022784233093262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,8,power_law_1.2,0.9468799591064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,8,power_law_1.01,0.8352448463439941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,8,power_law_1.2,1.0294400215148927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,8,power_law_1.01,0.8820672035217285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,8,power_law_1.2,1.184870433807373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,8,power_law_1.01,0.9729023933410644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,8,1,power_law_1.2,1.3158207893371583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,8,power_law_1.2,0.6429567813873291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,8,power_law_1.01,1.0815872192382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,8,power_law_1.2,1.352780818939209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,8,power_law_1.2,0.682047986984253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,8,power_law_1.01,1.2770112037658692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,8,power_law_1.2,1.5720191955566407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,8,power_law_1.2,0.7359551906585693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,8,power_law_1.01,1.5118144035339356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,8,power_law_1.2,0.8620991706848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,8,power_law_1.2,1.865056037902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,8,1,balanced,0.4228959878285726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,8,1,balanced,0.42551998297373456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,8,1,balanced,0.4294613202412923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,8,power_law_1.01,1.9584127426147462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,8,1,balanced,0.4314719835917155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,8,1,balanced,0.4328800042470296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,8,power_law_1.2,2.4247360229492188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,8,1,balanced,0.4387199878692627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,8,power_law_1.01,2.3689855575561523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,8,power_law_1.01,0.5340415954589843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,8,power_law_1.2,2.918272018432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,8,power_law_1.2,0.9836544036865235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,8,power_law_1.01,3.3214336395263673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,8,1,power_law_1.01,1.0303615570068358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,8,power_law_1.2,1.199443244934082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,8,1,power_law_1.2,1.84454402923584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,8,power_law_1.2,3.9880001068115236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,8,power_law_1.2,1.3988479614257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,8,power_law_1.01,4.181094360351563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,8,power_law_1.2,1.8618240356445312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,8,power_law_1.2,2.3310016632080077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,8,1,power_law_1.2,0.92674560546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,8,power_law_1.01,6.112416076660156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,8,power_law_1.2,3.2193153381347654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,8,power_law_1.01,0.5446208000183106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,8,power_law_1.2,5.151654434204102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,8,power_law_1.01,0.5834432125091553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,8,power_law_1.01,0.5899839878082276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,8,power_law_1.2,4.211372756958008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,8,power_law_1.01,0.6240831851959229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,8,power_law_1.01,0.6638591766357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,8,power_law_1.01,7.98221435546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,8,power_law_1.01,0.7199679851531983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,8,power_law_1.2,7.265459442138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,8,1,power_law_1.2,2.369087982177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,8,power_law_1.2,0.16843520402908324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,8,power_law_1.2,0.21402239799499512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,8,power_law_1.2,0.2399679899215698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,8,power_law_1.01,10.997689819335937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,8,power_law_1.2,0.29614078998565674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,8,power_law_1.2,0.34199678897857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,8,power_law_1.2,10.156256103515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,8,power_law_1.2,0.38339838981628416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,8,1,balanced,0.07041066884994507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,8,1,balanced,0.10274133086204529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,8,1,balanced,0.17087999979654947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,8,1,balanced,0.3017706672350566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,8,power_law_1.2,0.35798399448394774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,8,power_law_1.01,0.8453568458557129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,8,power_law_1.2,0.36277759075164795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,8,power_law_1.01,0.9444352149963379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,8,power_law_1.2,0.707539176940918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,8,power_law_1.01,1.1548288345336915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,8,power_law_1.2,0.7313600063323975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,8,power_law_1.2,0.7629631996154785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,8,1,balanced,0.558458685874939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,8,power_law_1.2,5.873030471801758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,8,power_law_1.2,0.7678847789764405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,8,1,balanced,0.7971093654632568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,8,1,balanced,0.8008800347646078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,8,power_law_1.2,0.8280960083007812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,8,1,power_law_1.01,0.3856192111968994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,8,1,balanced,0.8003626664479574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,8,power_law_1.01,14.523802185058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,8,1,balanced,0.7498133182525635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,8,1,power_law_1.01,0.392793607711792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,8,power_law_1.2,0.862060832977295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,8,1,balanced,0.755242665608724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,8,1,balanced,0.041120000183582306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,8,1,balanced,0.06402666866779327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,8,1,balanced,0.10116799672444661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,8,1,balanced,0.7801547050476074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,8,1,balanced,0.1711626648902893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,8,1,balanced,0.30027733246485394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,8,power_law_1.01,1.3651647567749023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,8,1,balanced,0.776367982228597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,8,1,balanced,0.4226826826731364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,8,1,balanced,0.7845760186513265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,8,1,balanced,0.7962559858957926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,8,1,balanced,0.8068426450093588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,8,power_law_1.01,1.8268928527832031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,8,1,balanced,0.818336009979248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,8,power_law_1.2,7.569670104980469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,8,1,power_law_1.2,0.3916287899017334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,8,1,balanced,0.8420053323109945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,8,1,balanced,1.1042773723602295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,8,1,power_law_1.2,0.40460801124572754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,8,1,power_law_1.01,0.38955519199371336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,8,1,balanced,1.0776480038960774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,8,1,power_law_1.01,0.05418879985809326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,8,1,balanced,1.1150720119476318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,8,power_law_1.01,2.243974494934082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,8,1,power_law_1.2,0.42211198806762695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,8,1,power_law_1.01,0.4287680149078369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,8,1,balanced,1.116037368774414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,8,power_law_1.2,0.9279616355895997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,8,1,balanced,1.1411840120951335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,8,1,power_law_1.2,0.42992000579833983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,8,1,power_law_1.01,0.09807999730110169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,8,1,power_law_1.01,0.4368192195892334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,8,1,balanced,1.194709300994873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,8,1,balanced,1.1640106836954753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,8,1,power_law_1.2,0.4473087787628174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,8,1,power_law_1.01,0.4672704219818115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,8,1,power_law_1.01,0.1660863995552063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,8,power_law_1.2,1.0013824462890626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,8,1,balanced,1.277728001276652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,8,1,power_law_1.2,0.47363839149475095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,8,1,balanced,2.0550880432128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,8,1,power_law_1.01,0.27180159091949463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,8,power_law_1.2,1.1125951766967774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,8,1,balanced,2.039093335469564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,8,1,balanced,2.100149313608805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,8,1,balanced,2.1962879498799643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,8,power_law_1.2,1.3300095558166505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,8,1,balanced,0.513808012008667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,8,1,balanced,0.5147466659545898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,8,1,balanced,0.48419201374053955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,8,power_law_1.2,1.5274111747741699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,8,1,balanced,0.4857493241628011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,8,1,balanced,3.2374773025512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,8,1,balanced,0.4954933325449626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,8,power_law_1.01,3.065056037902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,8,1,balanced,0.4905546506245931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,8,1,balanced,0.4998240073521932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,8,1,balanced,0.5109173456827799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,8,1,power_law_1.01,0.3717504024505615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,8,power_law_1.2,2.0686208724975588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,8,1,balanced,0.5213493506113688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,8,1,balanced,3.5818878809611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,8,1,balanced,0.5339626471201578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,8,1,balanced,0.5576800107955933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,8,1,power_law_1.01,0.6058176040649415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,8,1,balanced,0.7382880051930746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,8,1,balanced,2.967514673868815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,8,power_law_1.01,3.9312511444091798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,8,power_law_1.2,2.4695871353149412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,8,1,power_law_1.01,0.6443840026855469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,8,1,power_law_1.01,0.6578944206237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,8,1,balanced,3.098112106323242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,8,1,balanced,5.567445119222005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,8,1,power_law_1.01,0.6492032051086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,8,power_law_1.2,3.562944030761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,8,1,power_law_1.01,0.657254409790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,8,1,balanced,4.992474555969238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,8,1,power_law_1.01,0.7158912181854248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,8,1,balanced,0.7793386777242025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,8,1,balanced,0.8628693421681722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,8,1,power_law_1.01,0.7155712127685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,8,1,balanced,6.893978754679362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,8,1,balanced,0.9443039894104004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,8,1,power_law_1.01,0.72641921043396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,8,power_law_1.2,4.339039993286133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,8,1,balanced,1.4921919504801433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,8,1,power_law_1.01,0.04023680090904236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,8,1,power_law_1.01,0.7550208091735839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,8,1,balanced,6.114847819010417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,8,1,balanced,1.6397387186686199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,8,1,power_law_1.01,0.06240000128746033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,8,1,power_law_1.01,0.7627520084381103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,8,1,power_law_1.01,0.1005247950553894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,8,1,power_law_1.01,0.779366397857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,8,1,balanced,2.3510185877482095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,8,1,power_law_1.01,0.15932799577713014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,8,1,power_law_1.01,0.8427840232849121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,8,1,power_law_1.01,0.21335039138793946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,8,1,balanced,2.6898345947265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,8,power_law_1.2,6.222419357299804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,8,1,power_law_1.01,1.0727807998657226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,8,1,power_law_1.01,0.3301503896713257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,8,1,power_law_1.01,0.4128896236419678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,8,1,power_law_1.01,1.158944034576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,8,1,balanced,4.161322593688965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,8,1,power_law_1.01,0.42969598770141604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,8,1,power_law_1.01,0.40965118408203127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,8,1,power_law_1.01,0.41818881034851074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,8,power_law_1.01,5.752979278564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,8,1,power_law_1.01,0.4418623924255371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,8,1,balanced,5.177296002705892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,8,1,power_law_1.01,0.4509376049041748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,8,power_law_1.2,8.071949005126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,8,1,power_law_1.01,0.4664639949798584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,8,1,power_law_1.2,0.05489280223846436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,8,1,power_law_1.2,0.03930239975452423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,8,1,power_law_1.01,0.47905921936035156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,8,1,power_law_1.2,0.09743360280990601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,8,1,power_law_1.01,1.3513152122497558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,8,1,power_law_1.01,0.4999680042266846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,8,1,power_law_1.2,0.16563199758529662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,8,1,power_law_1.01,0.515231990814209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,8,1,power_law_1.01,1.5072768211364747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,8,1,power_law_1.01,0.5586368083953858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,8,1,power_law_1.01,0.7379712104797364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,8,power_law_1.01,7.470867156982422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,8,1,power_law_1.01,1.8598592758178711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,16,1,balanced,0.027376001079877216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,16,1,balanced,0.03345066557327906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,16,1,balanced,0.05479466418425242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,16,1,balanced,0.053727999329566956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,8,1,power_law_1.01,0.8138367652893066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,16,1,balanced,0.055498664577802025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,16,1,balanced,0.05491200089454651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,8,1,power_law_1.2,0.06079999804496765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,16,1,balanced,0.058058664202690125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,16,1,balanced,0.06018133461475372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,16,1,balanced,0.07302933434645335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,16,1,balanced,0.07447466750939687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,16,1,balanced,0.07311999797821045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,8,1,power_law_1.01,0.9511743545532226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,16,1,balanced,0.07648000121116638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,8,1,power_law_1.01,2.242361640930176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,16,1,balanced,0.07645333309968312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,16,1,balanced,0.07788266738255818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,16,1,balanced,0.09321066737174988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,16,1,balanced,0.093231995900472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,16,1,balanced,0.08613866567611694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,16,1,balanced,0.09711999694506328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,8,1,power_law_1.01,1.0918144226074218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,16,1,balanced,0.10104533036549886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,16,1,balanced,0.15133866667747498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,16,1,balanced,0.1710933248202006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,16,1,balanced,0.23706134160359701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,8,power_law_1.2,11.8499267578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,16,1,balanced,0.30004799365997314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,8,1,power_law_1.01,1.380076789855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,8,1,power_law_1.01,2.987500762939453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,8,1,power_law_1.2,0.09915519952774048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,8,1,power_law_1.2,0.244051194190979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,8,1,power_law_1.2,0.15152000188827514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,8,1,power_law_1.01,3.728985595703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,8,1,power_law_1.2,0.34283521175384524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,8,1,power_law_1.2,0.5965568065643311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,8,1,power_law_1.01,1.660691261291504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,8,1,power_law_1.01,1.1325823783874511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,8,1,power_law_1.2,1.1732799530029296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,8,1,power_law_1.01,1.2342144012451173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,8,1,power_law_1.01,5.200384140014648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,8,1,power_law_1.01,2.2315839767456054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,8,1,power_law_1.2,0.19912960529327392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,8,1,power_law_1.2,1.2729855537414552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,8,1,power_law_1.01,1.3888640403747559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,8,1,power_law_1.2,0.3275968074798584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,8,1,power_law_1.2,1.4249024391174316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,8,1,power_law_1.01,1.5539392471313476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,8,1,power_law_1.2,0.4146240234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,8,1,power_law_1.01,2.8175424575805663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,8,power_law_1.2,15.43939208984375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,8,1,power_law_1.2,1.618239974975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,16,1,balanced,0.02548266698916753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,16,1,balanced,0.029482667644818623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,16,1,balanced,0.03669866671164831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,16,1,balanced,0.03573866685231527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,16,1,balanced,0.035274667044480644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,16,1,balanced,0.03746666759252548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,8,1,power_law_1.01,1.9239551544189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,16,1,balanced,0.04502933224042257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,16,1,balanced,0.050069332122802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,16,1,balanced,0.0517493337392807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,16,1,balanced,0.05205333232879639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,16,1,balanced,0.04868266483147939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,8,1,power_law_1.2,1.9879039764404296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,16,1,balanced,0.04897066454092661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,8,1,power_law_1.01,6.759526062011719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,16,1,balanced,0.04994133114814758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,16,1,balanced,0.05392533540725708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,16,1,balanced,0.06975999971230824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,16,1,balanced,0.0707946668068568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,16,1,balanced,0.06618666648864746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,8,1,power_law_1.01,2.2510719299316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,16,1,balanced,0.0772266685962677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,16,1,balanced,0.07904533545176189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,16,1,balanced,0.11485866705576579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,16,1,balanced,0.12993066509564719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,8,1,power_law_1.2,0.6449984073638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,16,1,balanced,0.1790613333384196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,16,1,balanced,0.22417600949605307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,8,1,power_law_1.2,0.436191987991333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,16,1,balanced,0.3035306731859843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,8,1,power_law_1.2,0.655244779586792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,16,1,balanced,0.3901653289794922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,8,1,power_law_1.2,0.41736321449279784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,16,1,balanced,0.403600017229716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,16,1,balanced,0.5519786675771078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,8,1,power_law_1.01,2.9037567138671876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,8,1,power_law_1.2,0.4284031867980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,16,1,balanced,0.5245279868443807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,16,1,balanced,0.7528586387634277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,16,1,balanced,0.04178133110205332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,8,1,power_law_1.2,0.4487040042877197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,16,1,balanced,0.04569066564242045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,16,1,balanced,0.052000001072883606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,8,1,power_law_1.01,4.014559936523438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,16,1,balanced,0.05309866865475973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,16,1,balanced,0.9886186917622884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,16,1,balanced,0.051728000243504844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,16,1,balanced,0.054101333022117615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,8,1,power_law_1.2,0.4628287792205811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,16,1,balanced,0.05407466491063436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,16,1,balanced,0.05537599821885427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,8,1,power_law_1.2,2.2949951171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,16,1,balanced,0.07083199918270111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,16,1,balanced,0.07332266867160797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,8,1,power_law_1.2,0.4723328113555908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,16,1,balanced,0.07282666862010956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,8,1,power_law_1.01,3.6774974822998048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,16,1,balanced,0.07427733143170674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,16,1,balanced,0.08218133449554443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,16,1,balanced,0.08216000099976857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,16,1,balanced,0.0978613297144572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,16,1,balanced,0.09690133730570476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,8,1,power_law_1.2,0.4952703952789307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,16,1,balanced,0.0974720021088918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,16,1,balanced,0.14435199896494547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,16,1,balanced,0.16063466668128967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,16,1,balanced,0.2251946727434794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,16,1,balanced,0.7265653610229492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,8,1,power_law_1.2,3.034239959716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,16,1,balanced,0.280511995156606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,16,1,power_law_1.01,0.028812798857688903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,16,1,balanced,0.376911997795105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,16,1,power_law_1.01,0.034790399670600894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,16,1,balanced,0.48096001148223877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,8,1,power_law_1.01,5.180140686035156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,16,1,power_law_1.01,0.03774079978466034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,16,1,balanced,0.6841013431549072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,16,1,power_law_1.01,0.051641601324081424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,16,1,power_law_1.01,0.0244159996509552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,8,1,power_law_1.01,5.101433563232422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,16,1,power_law_1.01,0.05240319967269898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,16,1,power_law_1.01,0.03059200048446655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,16,1,balanced,0.9010132948557535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,16,1,power_law_1.01,0.05489919781684875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,16,1,power_law_1.01,0.031142398715019226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,16,1,power_law_1.01,0.056953597068786624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,16,1,power_law_1.01,0.036134400963783266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,16,1,balanced,0.025173333783944447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,16,1,power_law_1.01,0.06340479850769043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,16,1,balanced,0.025290665527184803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,16,1,power_law_1.01,0.03537279963493347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,16,1,balanced,0.029205332199732464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,16,1,power_law_1.01,0.07370240092277527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,16,1,balanced,0.029157333076000214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,16,1,power_law_1.01,0.039628800749778745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,16,1,power_law_1.01,0.07505919933319091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,16,1,power_law_1.01,0.05000320076942444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,16,1,power_law_1.01,0.07301120162010193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,8,1,power_law_1.2,0.5179776191711426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,16,1,power_law_1.01,0.05306239724159241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,16,1,power_law_1.01,0.07603840231895446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,8,1,power_law_1.2,0.6440447807312012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,16,1,power_law_1.01,0.05543680191040039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,16,1,power_law_1.01,0.07725440263748169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,16,1,power_law_1.01,0.06106879711151123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,8,1,power_law_1.2,0.5409984111785888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,16,1,power_law_1.01,0.08480640053749085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,8,1,power_law_1.2,0.661836814880371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,16,1,power_law_1.01,0.05058559775352478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,16,1,power_law_1.01,0.055769598484039305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,8,1,power_law_1.2,0.5744063854217529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,8,1,power_law_1.01,6.547545623779297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,16,1,power_law_1.01,0.057580798864364624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,16,1,power_law_1.01,0.037459200620651244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,16,1,balanced,1.2790613174438477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,8,1,power_law_1.2,0.7687359809875488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,16,1,power_law_1.01,0.06157439947128296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,16,1,power_law_1.01,0.04669440090656281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,16,1,balanced,0.029765332738558452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,16,1,power_law_1.01,0.046112000942230225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,16,1,balanced,0.03050133337577184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,16,1,power_law_1.01,0.06832640171051026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,16,1,balanced,0.03329599897066752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,16,1,balanced,1.6980266571044922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,16,1,power_law_1.01,0.050393599271774295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,16,1,balanced,0.03327466547489166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,8,1,power_law_1.2,0.8408384323120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,16,1,balanced,0.03565866748491923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,16,1,balanced,0.03495466709136963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,16,1,balanced,0.03737599899371465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,16,1,balanced,0.03938666731119156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,16,1,balanced,0.04043200115362803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,16,1,balanced,0.04188266893227895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,16,1,power_law_1.01,0.07334399819374085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,16,1,balanced,0.05272533496220907
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,16,1,balanced,0.0499893327554067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,16,1,balanced,0.045754666129748024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,16,1,balanced,0.08101333181063335
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,16,1,balanced,0.06532266736030579
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,16,1,balanced,0.08349333206812541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,16,1,balanced,0.0988106628259023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,16,1,balanced,0.1274133324623108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,8,1,power_law_1.2,3.7458751678466795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,16,1,balanced,0.16061866283416748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,16,1,balanced,0.22542399168014526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,16,1,balanced,0.2949440081914266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,16,1,balanced,0.42159998416900635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,16,1,balanced,0.5469333330790201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,16,1,power_law_1.2,0.028492799401283263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,16,1,power_law_1.01,0.050367999076843264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,16,1,power_law_1.2,0.03260799944400787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,16,1,power_law_1.01,0.0531391978263855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,16,1,power_law_1.2,0.03850240111351013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,16,1,power_law_1.01,0.05264639854431152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,8,1,power_law_1.2,0.9824000358581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,16,1,power_law_1.2,0.05066879987716675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,16,1,power_law_1.01,0.055174398422241214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,16,1,power_law_1.2,0.05192959904670715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,16,1,power_law_1.01,0.07039999961853027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,16,1,power_law_1.01,0.09011840224266052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,8,1,power_law_1.2,5.215039825439453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,16,1,power_law_1.01,0.06632320284843445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,8,1,power_law_1.2,1.1260224342346192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,16,1,power_law_1.01,0.07093759775161743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,16,1,power_law_1.01,0.09466879963874816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,16,1,power_law_1.01,0.07440639734268188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,16,1,power_law_1.01,0.08718720078468323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,16,1,power_law_1.01,0.08168320059776306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,16,1,power_law_1.01,0.09760640263557434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,8,1,power_law_1.2,1.4088831901550294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,16,1,power_law_1.01,0.08574720025062561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,16,1,power_law_1.01,0.0913536012172699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,16,1,power_law_1.01,0.09813759922981262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,8,1,power_law_1.2,1.6957632064819337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,8,1,power_law_1.2,0.7130176067352295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,16,1,power_law_1.01,0.1147007942199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,16,1,power_law_1.01,0.10174720287322998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,16,1,power_law_1.2,0.054176002740859985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,16,1,power_law_1.01,0.07591040134429931
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,16,1,power_law_1.01,0.026156800985336303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,16,1,power_law_1.2,0.05661439895629883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,16,1,power_law_1.01,0.02632319927215576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,16,1,power_law_1.2,0.05772799849510193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,16,1,power_law_1.01,0.10974719524383544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,16,1,power_law_1.01,0.027091199159622194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,16,1,power_law_1.2,0.07319679856300354
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,16,1,power_law_1.01,0.028652799129486085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,8,1,power_law_1.2,2.2700927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,16,1,power_law_1.2,0.07443199753761291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,16,1,power_law_1.01,0.029369598627090453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,16,1,power_law_1.2,0.07326719760894776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,16,1,power_law_1.01,0.08513919711112976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,16,1,power_law_1.2,0.07567999958992004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,16,1,power_law_1.2,0.0775551974773407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,16,1,power_law_1.01,0.1614464044570923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,16,1,power_law_1.2,0.07923840284347534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,8,1,power_law_1.2,2.8525568008422852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,16,1,power_law_1.2,0.09038720130920411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,16,1,power_law_1.2,0.09363840222358703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,16,1,power_law_1.01,0.03144319951534271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,16,1,power_law_1.01,0.03270399868488312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,16,1,power_law_1.01,0.15178879499435424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,16,1,power_law_1.01,0.034272000193595886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,16,1,power_law_1.01,0.034380799531936644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,16,1,power_law_1.2,0.0874559998512268
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,16,1,power_law_1.01,0.03681280016899109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,16,1,power_law_1.01,0.17904000282287597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,16,1,power_law_1.01,0.03666560053825378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,16,1,power_law_1.01,0.11932799816131592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,16,1,power_law_1.01,0.03930880129337311
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,16,1,power_law_1.01,0.040992000699043275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,16,1,power_law_1.2,0.09742720127105713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,16,1,power_law_1.01,0.042982399463653564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,8,1,power_law_1.2,0.7262464046478272
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,16,1,power_law_1.01,0.043507200479507444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,16,1,power_law_1.01,0.16547839641571044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,16,1,power_law_1.01,0.04578559994697571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,8,1,power_law_1.2,0.7421055793762207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,16,1,power_law_1.2,0.10936959981918334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,16,1,power_law_1.01,0.048076799511909483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,8,1,power_law_1.2,0.7666687965393066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,16,1,power_law_1.01,0.13506560325622557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,8,1,power_law_1.2,0.8028032302856445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,8,1,power_law_1.2,4.072544097900391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,16,1,power_law_1.2,0.1604416012763977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,8,1,power_law_1.2,0.8285568237304688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,16,1,power_law_1.01,0.06120319962501526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,16,1,power_law_1.01,0.24522240161895753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,8,1,power_law_1.2,0.8636096000671387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,16,1,power_law_1.01,0.22032639980316163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,16,1,power_law_1.2,0.1780992031097412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,16,1,power_law_1.01,0.06971520185470581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,16,1,power_law_1.01,0.2825344085693359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,16,1,power_law_1.01,0.08389120101928711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,8,1,power_law_1.2,5.24835205078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,16,1,power_law_1.2,0.024607999622821806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,16,1,power_law_1.2,0.029849600791931153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,16,1,power_law_1.2,0.02991360127925873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,16,1,power_law_1.01,0.09902719855308532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,16,1,power_law_1.01,0.18516479730606078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,16,1,power_law_1.2,0.03466239869594574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,16,1,power_law_1.2,0.037452799081802365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,16,1,power_law_1.01,0.3091775894165039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,16,1,power_law_1.2,0.0409855991601944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,8,1,power_law_1.2,1.120473575592041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,16,1,power_law_1.2,0.04759039878845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,16,1,power_law_1.2,0.0525056004524231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,16,1,power_law_1.2,0.05886080265045166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,8,1,power_law_1.2,1.2047103881835937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,16,1,power_law_1.2,0.05985919833183288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,16,1,power_law_1.2,0.05237759947776795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,16,1,power_law_1.2,0.05627520084381103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,8,1,power_law_1.2,1.3759360313415527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,16,1,power_law_1.01,0.2278143882751465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,16,1,power_law_1.2,0.05786240100860596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,16,1,power_law_1.01,0.12635519504547119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,16,1,power_law_1.2,0.2408384084701538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,8,1,power_law_1.2,1.5551103591918944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,16,1,power_law_1.2,0.06289920210838318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,16,1,power_law_1.2,0.06992639899253845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,16,1,power_law_1.01,0.3101183891296387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,16,1,power_law_1.2,0.0751039981842041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,8,1,power_law_1.2,1.9154367446899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,16,1,power_law_1.01,0.3743551969528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,16,1,power_law_1.2,0.06647679805755616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,8,1,power_law_1.2,2.2778112411499025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,16,1,power_law_1.01,0.16081279516220093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,16,1,power_law_1.2,0.07624959945678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,8,1,power_law_1.2,3.0100992202758787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,16,1,power_law_1.2,0.3047935962677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,8,1,power_law_1.2,3.7323326110839843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,16,1,power_law_1.2,0.09181439876556396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,16,1,power_law_1.01,0.4142655849456787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,8,1,power_law_1.2,6.635001373291016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,8,1,power_law_1.2,5.228031921386719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,16,1,power_law_1.2,0.03607679903507233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,16,1,power_law_1.2,0.040582400560379026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,16,1,power_law_1.2,0.12009600400924683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,16,1,power_law_1.2,0.04679040014743805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,16,1,power_law_1.2,0.049881601333618165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,16,1,power_law_1.2,0.04933759868144989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,16,1,power_law_1.2,0.052742397785186766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,16,1,power_law_1.2,0.053011202812194826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,16,1,power_law_1.01,0.47888641357421874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,16,1,power_law_1.2,0.055904000997543335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,16,1,power_law_1.2,0.0713599979877472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,16,1,power_law_1.2,0.07314559817314148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,16,1,power_law_1.2,0.0796288013458252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,16,1,power_law_1.2,0.0841152012348175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,8,1,power_law_1.2,6.924588775634765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,16,1,power_law_1.2,0.08674560189247131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,16,1,power_law_1.01,0.40079360008239745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,16,1,power_law_1.2,0.09147520065307617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,16,1,power_law_1.2,0.0972927987575531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,16,1,power_law_1.2,0.10163840055465698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,16,1,power_law_1.2,0.10304640531539917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,16,1,power_law_1.2,0.025139200687408447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,16,1,power_law_1.2,0.15252480506896973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,16,1,power_law_1.2,0.02571519911289215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,16,1,power_law_1.2,0.02678399980068207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,16,1,power_law_1.2,0.028191998600959778
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,16,1,power_law_1.2,0.02876160144805908
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,16,1,power_law_1.2,0.03125759959220886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,16,1,power_law_1.2,0.032492798566818235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,16,1,power_law_1.2,0.03368319869041443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,16,1,power_law_1.2,0.03449600040912628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,16,1,power_law_1.2,0.1364416003227234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,16,1,power_law_1.2,0.034969601035118106
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,16,1,power_law_1.2,0.035411199927330016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,16,1,power_law_1.2,0.03806079924106598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,16,1,power_law_1.2,0.16458239555358886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,16,1,power_law_1.2,0.04007680118083954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,16,1,power_law_1.2,0.042444801330566405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,16,1,power_law_1.2,0.04289279878139496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,16,1,power_law_1.2,0.0455487996339798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,16,1,power_law_1.2,0.1810688018798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,16,1,power_law_1.2,0.22349441051483154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,16,1,power_law_1.2,0.04745599925518036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,16,1,power_law_1.01,0.22668159008026123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,16,1,power_law_1.2,0.059334397315979004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,16,1,power_law_1.2,0.0670144021511078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,16,1,power_law_1.2,0.41235837936401365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,16,1,power_law_1.2,0.22536959648132324
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,16,1,power_law_1.2,0.08264319896697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,16,1,power_law_1.01,0.6737599849700928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,16,1,power_law_1.2,0.28706560134887693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,16,1,power_law_1.01,0.5734208106994629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,16,1,power_law_1.01,0.7441728115081787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,16,1,balanced,0.03346133232116699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,16,1,balanced,0.049679999550183616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,16,1,balanced,0.07106666763623555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,16,1,balanced,0.07203733424345653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,16,1,balanced,0.07085866729418437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,16,1,balanced,0.07222933570543925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,16,1,balanced,0.08345066507657369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,16,1,balanced,0.08377066254615784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,16,1,balanced,0.1006666620572408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,16,1,balanced,0.10317333539326985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,16,1,balanced,0.1014400025208791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,16,1,balanced,0.10358933607737224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,16,1,balanced,0.10291199882825215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,16,1,balanced,0.10803733269373576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,16,1,balanced,0.1220693290233612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,16,1,balanced,0.12333866953849792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,16,1,balanced,0.11690666278203328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,16,1,balanced,0.13487999637921652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,16,1,balanced,0.14201600352923074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,16,1,balanced,0.21798932552337646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,16,1,balanced,0.2458560069402059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,16,1,balanced,0.34540267785390216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,16,1,balanced,0.44649600982666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,16,1,balanced,0.6461600065231323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,16,1,balanced,0.8174026807149252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,16,1,power_law_1.2,0.09872639775276185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,16,1,balanced,1.2100586891174316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,16,1,balanced,1.579253355662028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,16,1,balanced,0.0314026673634847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,16,1,balanced,0.04109866668780645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,16,1,balanced,0.05807999769846598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,16,1,balanced,0.056128000219662987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,16,1,balanced,0.056405335664749146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,16,1,balanced,0.05972800155480703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,16,1,balanced,0.07218666871388753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,16,1,balanced,0.07833600044250488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,16,1,balanced,0.08637866377830505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,16,1,balanced,0.08458133538564046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,16,1,balanced,0.0713919997215271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,16,1,balanced,0.07467733323574066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,16,1,balanced,0.0786293347676595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,16,1,balanced,0.07879466811815898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,16,1,balanced,0.09061333537101746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,16,1,balanced,0.0972053309281667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,16,1,balanced,0.08682133754094441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,16,1,balanced,0.10531199971834819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,16,1,balanced,0.11010133226712544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,16,1,balanced,0.160261332988739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,16,1,balanced,0.18338133891423544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,16,1,balanced,0.2539520064989726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,16,1,balanced,0.32234134276707965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,16,1,balanced,0.4539946715037028
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,16,1,power_law_1.01,0.295084810256958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,16,1,balanced,0.582202672958374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,16,1,balanced,0.8399679660797119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,16,1,balanced,1.100501298904419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,16,1,balanced,0.04566933214664459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,16,1,balanced,0.059994667768478394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,16,1,balanced,0.0672106643517812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,16,1,balanced,0.06844266752401988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,16,1,balanced,0.06834666430950165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,16,1,balanced,0.07451733450094859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,16,1,balanced,0.07568533221880595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,16,1,balanced,0.0746559997399648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,16,1,balanced,0.10865599910418193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,16,1,balanced,0.10962667067845662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,16,1,balanced,0.10866666833559673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,16,1,balanced,0.11473600069681804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,16,1,balanced,0.11469866832097371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,16,1,balanced,0.12298666437466939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,16,1,balanced,0.13729066650072733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,16,1,balanced,0.1395146648089091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,16,1,power_law_1.01,0.5381760120391845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,16,1,balanced,0.1346826652685801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,16,1,power_law_1.2,0.5385791778564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,16,1,balanced,0.206661323706309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,16,1,balanced,0.22989332675933838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,16,1,balanced,0.3227199912071228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,16,1,balanced,0.4162293275197347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,16,1,balanced,0.5946400165557861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,16,1,balanced,0.74507737159729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,16,1,balanced,1.088378667831421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,16,1,power_law_1.2,0.37176320552825926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,16,1,balanced,1.4229547182718914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,16,1,balanced,2.0347572962443032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,16,1,power_law_1.01,0.8788928031921387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,16,1,balanced,2.6618453661600747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,16,1,power_law_1.2,0.12910079956054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,16,1,balanced,0.026234666506449383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,16,1,balanced,0.029178666571776073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,16,1,balanced,0.0315786674618721
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,16,1,balanced,0.03169599920511246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,16,1,balanced,0.03266666581233343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,16,1,balanced,0.033413333197434746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,16,1,balanced,0.04211199780305227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,16,1,balanced,0.04390400151411692
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,16,1,balanced,0.044490665197372437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,16,1,balanced,0.04457066456476847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,16,1,balanced,0.045925334095954895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,16,1,balanced,0.04991999765237173
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,16,1,balanced,0.0505973349014918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,16,1,balanced,0.05751466751098633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,16,1,balanced,0.06939733525117238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,16,1,balanced,0.06549333532651265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,16,1,balanced,0.0646613339583079
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,16,1,balanced,0.10361066460609436
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,16,1,balanced,0.0846613347530365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,16,1,balanced,0.10471466183662415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,16,1,balanced,0.12935466567675272
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,16,1,balanced,0.17756799856821695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,16,1,balanced,0.2234826683998108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,16,1,balanced,0.32392533620198566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,16,1,power_law_1.01,0.7736447811126709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,16,1,power_law_1.2,0.4843455791473389
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,16,1,balanced,0.4161493380864461
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,16,1,balanced,0.6010933319727579
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,16,1,balanced,0.7842933336893717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,16,1,power_law_1.01,0.03831680119037628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,16,1,power_law_1.01,0.04636160135269165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,16,1,power_law_1.01,0.055264002084732054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,16,1,power_law_1.01,0.06973440051078797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,16,1,power_law_1.01,0.06964480280876159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,16,1,power_law_1.01,0.07412480115890503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,16,1,power_law_1.01,0.08187519907951354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,16,1,power_law_1.01,0.08967679738998413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,16,1,power_law_1.01,0.10236159563064576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,16,1,power_law_1.01,0.10317440032958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,16,1,power_law_1.01,0.10643839836120605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,16,1,power_law_1.01,0.10645120143890381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,16,1,power_law_1.01,0.11185920238494873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,16,1,power_law_1.01,0.11527680158615113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,16,1,power_law_1.01,0.12037760019302368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,16,1,power_law_1.01,0.12348159551620483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,16,1,power_law_1.01,0.12729599475860595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,16,1,power_law_1.01,0.13955199718475342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,16,1,power_law_1.01,0.19304959774017333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,16,1,power_law_1.2,0.3083775997161865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,16,1,power_law_1.2,0.1602944016456604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,16,1,power_law_1.01,0.22611839771270753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,16,1,power_law_1.01,0.42356481552124026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,16,1,power_law_1.01,0.25880320072174073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,16,1,power_law_1.2,0.7738880157470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,16,1,power_law_1.01,1.2720767974853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,16,1,power_law_1.01,0.35797119140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,16,1,power_law_1.2,0.6858816146850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,16,1,power_law_1.2,0.2257983922958374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,16,1,power_law_1.01,0.5518208026885987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,16,1,power_law_1.01,0.036499199271202085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,16,1,power_law_1.01,0.040627199411392215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,16,1,power_law_1.01,0.04589439928531647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,16,1,power_law_1.01,0.05374079942703247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,16,1,power_law_1.01,0.053887999057769774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,16,1,power_law_1.01,0.05734400153160095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,16,1,power_law_1.01,0.06810240149497986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,16,1,power_law_1.01,0.07292159795761108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,16,1,power_law_1.01,0.0828224003314972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,16,1,power_law_1.01,0.08410239815711976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,16,1,power_law_1.01,0.07416319847106934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,16,1,power_law_1.01,0.07868800163269044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,16,1,power_law_1.01,0.08061439990997314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,16,1,power_law_1.01,0.08332800269126892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,16,1,power_law_1.01,0.4607423782348633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,16,1,power_law_1.01,0.08895999789237977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,16,1,power_law_1.01,0.09342079758644103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,16,1,power_law_1.01,0.09221119880676269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,16,1,power_law_1.01,0.10426239967346192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,16,1,power_law_1.01,0.13656320571899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,16,1,power_law_1.01,0.16650240421295165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,16,1,power_law_1.2,0.3913664102554321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,16,1,power_law_1.01,0.18918399810791015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,16,1,power_law_1.01,1.6738367080688477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,16,1,power_law_1.01,0.0554751992225647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,16,1,power_law_1.01,0.05925760269165039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,16,1,power_law_1.01,0.06213120222091675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,16,1,power_law_1.01,0.06728320121765137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,16,1,power_law_1.01,0.06771199703216553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,16,1,power_law_1.01,0.07191039919853211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,16,1,power_law_1.01,0.07289599776268005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,16,1,power_law_1.01,0.07514879703521729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,16,1,power_law_1.01,0.1066431999206543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,16,1,power_law_1.01,0.2606208086013794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,16,1,power_law_1.01,0.10792319774627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,16,1,power_law_1.01,0.1096127986907959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,16,1,power_law_1.01,0.11331839561462402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,16,1,power_law_1.01,0.11448960304260254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,16,1,power_law_1.01,0.6631423950195312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,16,1,power_law_1.01,0.12475520372390747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,16,1,power_law_1.01,0.13509759902954102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,16,1,power_law_1.01,0.1386240005493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,16,1,power_law_1.01,0.1893504023551941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,16,1,power_law_1.01,0.21244161128997802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,16,1,power_law_1.01,0.24706559181213378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,16,1,power_law_1.2,0.2946367979049683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,16,1,power_law_1.01,0.32999041080474856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,16,1,power_law_1.01,0.32746880054473876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,16,1,power_law_1.01,0.4201344013214111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,16,1,power_law_1.01,0.8543359756469726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,16,1,power_law_1.01,0.5826432228088378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,16,1,power_law_1.01,1.0123071670532227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,16,1,power_law_1.01,0.7374015808105469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,16,1,power_law_1.2,1.0133567810058595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,16,1,power_law_1.01,0.02481919974088669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,16,1,power_law_1.2,0.037625598907470706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,16,1,power_law_1.2,0.04516479969024658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,16,1,power_law_1.01,0.026144000887870788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,16,1,power_law_1.2,0.05271040201187134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,16,1,power_law_1.01,0.02741119861602783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,16,1,power_law_1.2,0.06951680183410644
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,16,1,power_law_1.01,0.030348798632621764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,16,1,power_law_1.2,0.0688256025314331
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,16,1,power_law_1.01,0.032051199674606325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,16,1,power_law_1.2,0.07425280213356018
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,16,1,power_law_1.01,0.037088000774383546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,16,1,power_law_1.2,0.0821951985359192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,16,1,power_law_1.01,0.4647871971130371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,16,1,power_law_1.2,0.09070720076560974
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,16,1,power_law_1.01,0.03973119854927063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,16,1,power_law_1.2,0.10246399641036988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,16,1,power_law_1.01,0.04481280148029328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,16,1,power_law_1.2,0.10405119657516479
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,16,1,power_law_1.01,0.041868799924850465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,16,1,power_law_1.2,0.1044160008430481
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,16,1,power_law_1.01,0.043935999274253845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,16,1,power_law_1.01,0.048716801404953006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,16,1,power_law_1.01,0.04936319887638092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,16,1,power_law_1.2,0.10953600406646728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,16,1,power_law_1.01,0.05050879716873169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,16,1,power_law_1.2,0.11084799766540528
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,16,1,power_law_1.01,0.058931201696395874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,16,1,power_law_1.2,0.11593600511550903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,16,1,power_law_1.01,0.05749760270118713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,16,1,power_law_1.01,0.059622400999069215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,16,1,power_law_1.01,0.06791679859161377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,16,1,power_law_1.2,0.8786687850952148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,16,1,power_law_1.01,0.07797759771347046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,16,1,power_law_1.2,0.11987199783325195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,16,1,power_law_1.01,0.0899392008781433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,16,1,power_law_1.01,0.11147520542144776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,16,1,power_law_1.2,0.12499200105667115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,16,1,power_law_1.2,0.5656447887420655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,16,1,power_law_1.2,0.12451839447021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,16,1,power_law_1.2,0.13931519985198976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,16,1,power_law_1.01,0.13222399950027466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,16,1,power_law_1.01,0.17959680557250976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,16,1,power_law_1.2,0.19372160434722902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,16,1,power_law_1.01,1.0612992286682128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,16,1,power_law_1.01,1.2312383651733398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,16,1,power_law_1.01,1.3658176422119142
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,16,1,power_law_1.01,0.23077120780944824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,16,1,power_law_1.2,0.22614400386810302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,16,1,power_law_1.01,0.5939072132110595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,16,1,power_law_1.2,0.738316822052002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,16,1,power_law_1.01,0.33061120510101316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,16,1,power_law_1.2,0.2679487943649292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,16,1,power_law_1.2,0.032876798510551454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,16,1,power_law_1.2,0.0403903990983963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,16,1,power_law_1.2,0.04346239864826203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,16,1,power_law_1.2,0.052262401580810545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,16,1,power_law_1.2,0.0536191999912262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,16,1,power_law_1.2,0.057843202352523805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,16,1,power_law_1.2,0.0701312005519867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,16,1,power_law_1.2,0.07214720249176025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,16,1,power_law_1.2,0.08318079710006714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,16,1,power_law_1.2,0.08533759713172913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,16,1,power_law_1.2,0.07096319794654846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,16,1,power_law_1.2,0.07751039862632751
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,16,1,power_law_1.2,0.4230527877807617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,16,1,power_law_1.2,0.07956479787826538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,16,1,power_law_1.2,0.08366720080375671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,16,1,power_law_1.01,0.4266047954559326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,16,1,power_law_1.2,0.08837760090827942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,16,1,power_law_1.2,0.3574975967407227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,16,1,power_law_1.2,0.0942911982536316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,16,1,power_law_1.2,0.08975359797477722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,16,1,power_law_1.2,0.10405119657516479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,16,1,power_law_1.01,1.996281623840332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,16,1,power_law_1.2,0.13526400327682495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,16,1,power_law_1.2,0.16472959518432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,16,1,power_law_1.01,1.6238143920898438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,16,1,power_law_1.2,0.05320320129394531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,16,1,power_law_1.2,0.05839999914169312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,16,1,power_law_1.2,0.060134398937225345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,16,1,power_law_1.2,0.06689280271530151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,16,1,power_law_1.2,0.06798719763755798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,16,1,power_law_1.2,0.07089920043945312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,16,1,power_law_1.2,0.07272959947586059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,16,1,power_law_1.2,0.1876479983329773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,16,1,power_law_1.2,0.07823359966278076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,16,1,power_law_1.2,0.10618879795074462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,16,1,power_law_1.2,0.10723199844360351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,16,1,power_law_1.01,0.8592320442199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,16,1,power_law_1.2,0.11094399690628051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,16,1,power_law_1.2,0.11219199895858764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,16,1,power_law_1.2,0.11509120464324951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,16,1,power_law_1.2,0.12640639543533325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,16,1,power_law_1.2,0.13518719673156737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,16,1,power_law_1.2,0.45862398147583006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,16,1,power_law_1.2,0.13928960561752318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,16,1,power_law_1.2,0.18973439931869507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,16,1,power_law_1.2,0.213702392578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,16,1,power_law_1.2,0.2380608081817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,16,1,power_law_1.2,0.25664639472961426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,16,1,power_law_1.2,0.32841598987579346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,16,1,power_law_1.01,2.630636787414551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,16,1,power_law_1.2,0.41994237899780273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,16,1,power_law_1.2,0.027923199534416198
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,16,1,power_law_1.2,0.029414400458335876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,16,1,power_law_1.2,0.029311999678611755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,16,1,power_law_1.2,0.032767999172210696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,16,1,power_law_1.2,0.032691198587417605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,16,1,power_law_1.2,0.03769600093364715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,16,1,power_law_1.2,0.04037759900093078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,16,1,power_law_1.2,0.04565759897232056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,16,1,power_law_1.2,0.04347519874572754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,16,1,power_law_1.2,0.04413439929485321
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,16,1,power_law_1.2,0.04888960123062134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,16,1,power_law_1.2,0.048742398619651794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,16,1,power_law_1.2,0.04940159916877747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,16,1,power_law_1.2,0.05914880037307739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,16,1,power_law_1.2,0.3206912040710449
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,16,1,power_law_1.2,0.05989760160446167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,16,1,power_law_1.2,1.278611183166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,16,1,power_law_1.2,0.06064000129699707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,16,1,power_law_1.2,0.06914560198783874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,16,1,power_law_1.2,0.6604544162750244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,16,1,power_law_1.2,0.0789247989654541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,16,1,power_law_1.2,0.09224320054054261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,16,1,power_law_1.2,0.5871551990509033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,16,1,power_law_1.01,0.6116672039031983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,16,1,power_law_1.2,0.45583357810974123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,16,1,power_law_1.2,0.11301120519638061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,16,1,power_law_1.2,0.13891839981079102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,16,1,power_law_1.01,1.1243776321411132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,16,balanced,0.051413332422574363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,16,balanced,0.05172266562779745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,16,balanced,0.06541866560777028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,16,balanced,0.09026133020718892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,16,balanced,0.16119999686876932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,16,balanced,0.2722880045572917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,16,balanced,0.26748265822728473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,16,balanced,0.26842667659123737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,16,balanced,0.27844266096750897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,16,balanced,0.2818613251050313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,16,balanced,0.36963733037312824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,16,balanced,0.3775093158086141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,16,balanced,0.38195733229319256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,16,balanced,0.39107199509938556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,16,balanced,0.40036265055338544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,16,balanced,0.4099466800689697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,16,balanced,0.42898666858673096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,16,balanced,0.4724746545155843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,16,balanced,0.5045653184254965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,16,1,power_law_1.2,0.1838655948638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,16,balanced,0.5672373374303182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,16,balanced,0.6277920007705688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,16,balanced,0.7545013427734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,16,balanced,0.8825600147247314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,16,balanced,1.2678986390431721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,16,balanced,1.5242506663004558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,16,balanced,2.1923680305480957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,16,1,power_law_1.2,0.7417024135589599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,16,balanced,2.857237180074056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,16,balanced,0.03990933299064636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,16,balanced,0.041834667325019836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,16,balanced,0.052095999320348106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,16,balanced,0.06852266689141591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,16,balanced,0.09834133585294087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,16,balanced,0.16792533795038858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,16,balanced,0.1941173275311788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,16,balanced,0.19584532578786215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,16,balanced,0.17732266585032144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,16,balanced,0.1844480037689209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,16,balanced,0.26158400376637775
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,16,1,power_law_1.2,0.23029119968414308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,16,balanced,0.27373333772023517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,16,balanced,0.28015466531117755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,16,balanced,0.29108800490697223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,16,balanced,0.303930660088857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,16,balanced,0.3160266677538554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,16,balanced,0.33691199620564777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,16,balanced,0.37595733006795246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,16,balanced,0.4093386729558309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,16,balanced,0.4779680172602336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,16,balanced,0.5482879877090454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,16,balanced,0.6833919684092203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,16,balanced,0.8190826574961344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,16,balanced,1.1662720044453938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,16,1,power_law_1.2,0.5496384143829346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,16,balanced,1.440000057220459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,16,balanced,0.0636053333679835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,16,balanced,0.06480533381303151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,16,balanced,0.07483733197053273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,16,balanced,0.08716266353925069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,16,balanced,0.10830400387446086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,16,balanced,0.16294933358828226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,16,balanced,0.1657919983069102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,16,balanced,2.07148806254069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,16,balanced,0.1683680017789205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,16,balanced,0.322650671005249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,16,balanced,0.3273973266283671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,16,balanced,0.33480000495910645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,16,balanced,0.34140264987945557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,16,balanced,2.685994784037272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,16,1,power_law_1.2,0.8536255836486817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,16,1,power_law_1.2,1.6721920013427733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,16,power_law_1.01,0.15470080375671386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,16,power_law_1.01,0.2456063985824585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,16,balanced,0.35115734736124676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,16,balanced,0.36319998900095624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,16,power_law_1.01,0.24686079025268554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,16,balanced,0.3797760009765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,16,balanced,0.39265600840250653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,16,1,power_law_1.2,1.0630208015441895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,16,power_law_1.01,0.2384768009185791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,16,balanced,0.4225653409957886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,16,balanced,0.473141352335612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,16,power_law_1.01,0.25345280170440676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,16,balanced,0.522815982500712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,16,power_law_1.01,0.18788479566574096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,16,balanced,0.6259466807047526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,16,balanced,0.7282773653666178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,16,1,power_law_1.2,0.5856063842773438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,16,balanced,1.074021339416504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,16,balanced,1.2799466451009114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,16,balanced,1.828565279642741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,16,balanced,2.3789119720458984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,16,power_law_1.01,0.24778239727020263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,16,balanced,0.03127466638882955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,16,balanced,0.03324266771475474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,16,balanced,0.03346133232116699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,16,balanced,0.04148799926042557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,16,power_law_1.01,0.257043194770813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,16,balanced,0.062496001521746315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,16,balanced,0.09301333626111348
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,16,balanced,0.09502399961153667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,16,power_law_1.01,0.24162559509277343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,16,balanced,0.09546666344006856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,16,balanced,0.09761599699656169
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,16,balanced,0.0981333355108897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,16,power_law_1.01,0.26282880306243894
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,16,balanced,0.09916266798973083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,16,balanced,0.13749333222707114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,16,balanced,0.13265066345532736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,16,power_law_1.01,0.3498624086380005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,16,balanced,0.1264959971110026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,16,balanced,0.29548267523447674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,16,power_law_1.01,0.33886079788208007
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,16,balanced,0.24184532960255942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,16,balanced,0.1904266675313314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,16,power_law_1.01,0.34641280174255373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,16,balanced,0.14453867077827454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,16,balanced,0.14524799585342407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,16,power_law_1.01,0.3620032072067261
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,16,balanced,0.1628320018450419
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,16,balanced,0.18238933881123862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,16,power_law_1.01,0.3659264087677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,16,balanced,0.22257065773010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,16,balanced,0.26782933870951336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,16,power_law_1.01,0.38817920684814455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,16,balanced,3.4756905237833657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,16,balanced,0.3779946565628052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,16,power_law_1.01,0.40955519676208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,16,balanced,0.463589350382487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,16,power_law_1.01,0.47571840286254885
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,16,balanced,0.6610826651255289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,16,power_law_1.01,0.5153728008270264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,16,balanced,0.8638666470845541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,16,balanced,4.571344057718913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,16,power_law_1.01,0.6132607936859131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,16,power_law_1.01,0.08871039748191833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,16,power_law_1.01,0.14809600114822388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,16,power_law_1.01,0.6912000179290771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,16,power_law_1.01,0.10029439926147461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,16,power_law_1.01,0.14755840301513673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,16,power_law_1.01,0.8964799880981446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,16,power_law_1.01,0.15518720149993898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,16,power_law_1.01,0.14880640506744386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,16,power_law_1.01,0.15557119846343995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,16,power_law_1.01,0.15004160404205322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,16,power_law_1.01,0.15845119953155518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,16,power_law_1.01,0.145907199382782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,16,power_law_1.01,0.1584831953048706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,16,power_law_1.01,0.16386560201644898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,16,power_law_1.01,0.13960319757461548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,16,power_law_1.01,0.16903040409088135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,16,power_law_1.01,0.15448960065841674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,16,power_law_1.01,0.16883840560913085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,16,power_law_1.01,0.1574079990386963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,16,power_law_1.01,0.16094720363616943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,16,power_law_1.01,0.3147775888442993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,16,power_law_1.01,0.23864960670471191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,16,power_law_1.01,1.0974335670471191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,16,power_law_1.01,0.26017920970916747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,16,power_law_1.01,0.31369600296020506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,16,power_law_1.01,0.26389760971069337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,16,power_law_1.01,0.3191359996795654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,16,power_law_1.01,1.4896320343017577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,16,power_law_1.01,0.2720128059387207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,16,power_law_1.01,0.32648959159851076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,16,power_law_1.01,0.2863744020462036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,16,power_law_1.01,0.3354496002197266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,16,power_law_1.01,0.294598388671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,16,power_law_1.01,1.834707260131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,16,power_law_1.01,0.33422079086303713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,16,power_law_1.01,0.3237375974655151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,16,power_law_1.01,0.3586496114730835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,16,power_law_1.01,0.3753727912902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,16,power_law_1.01,0.3907455921173096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,16,power_law_1.01,0.4150720119476318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,16,power_law_1.01,2.820800018310547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,16,power_law_1.01,0.41930880546569826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,16,power_law_1.01,0.5045760154724122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,16,power_law_1.01,0.5365312099456787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,16,power_law_1.01,0.5832384109497071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,16,power_law_1.01,0.602239990234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,16,power_law_1.01,0.7607232093811035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,16,power_law_1.01,0.7483391761779785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,16,power_law_1.01,3.717670440673828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,16,power_law_1.01,0.9604351997375489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,16,power_law_1.01,0.9217472076416016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,16,power_law_1.01,1.296268844604492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,16,power_law_1.01,1.2529151916503907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,16,1,power_law_1.2,1.2285504341125488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,16,power_law_1.01,1.6323968887329101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,16,power_law_1.01,1.5789504051208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,16,1,power_law_1.01,0.7930880069732666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,16,power_law_1.2,0.15291520357131957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,16,power_law_1.01,2.331955146789551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,16,power_law_1.2,0.24623360633850097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,16,power_law_1.01,2.3640127182006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,16,1,power_law_1.2,1.376799964904785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,16,power_law_1.2,0.19610879421234131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,16,power_law_1.2,0.2487488031387329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,16,1,power_law_1.2,0.32780799865722654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,16,power_law_1.2,0.2544703960418701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,16,power_law_1.01,3.083897590637207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,16,power_law_1.2,0.2515455961227417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,16,power_law_1.01,3.1540096282958983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,16,power_law_1.2,0.24912641048431397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,16,power_law_1.2,0.08796160221099854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,16,power_law_1.2,0.2571007966995239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,16,power_law_1.2,0.1487552046775818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,16,power_law_1.2,0.2479487895965576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,16,power_law_1.2,0.143449604511261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,16,power_law_1.2,0.2689728021621704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,16,power_law_1.2,0.14872959852218628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,16,power_law_1.2,0.3220736026763916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,16,power_law_1.2,0.1500607967376709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,16,power_law_1.01,4.336185455322266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,16,power_law_1.2,0.3695552110671997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,16,power_law_1.2,0.12621439695358277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,16,power_law_1.2,0.34163200855255127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,16,power_law_1.2,0.1779647946357727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,16,power_law_1.2,0.3606015920639038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,16,power_law_1.2,0.18565759658813477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,16,power_law_1.2,0.3773632049560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,16,power_law_1.2,0.158297598361969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,16,power_law_1.2,0.3924351930618286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,16,power_law_1.2,0.17493120431900025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,16,power_law_1.2,0.4109055995941162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,16,power_law_1.2,0.24106240272521973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,16,power_law_1.01,5.474668884277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,16,power_law_1.2,0.49515519142150877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,16,power_law_1.2,0.2557440042495728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,16,power_law_1.01,0.056524801254272464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,16,power_law_1.2,0.25127038955688474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,16,power_law_1.2,0.5413504123687745
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,16,power_law_1.01,0.08443520069122315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,16,power_law_1.2,0.09939200282096863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,16,power_law_1.2,0.2773184061050415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,16,power_law_1.2,0.6286399841308594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,16,power_law_1.2,0.1457152009010315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,16,power_law_1.01,0.08172799944877625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,16,power_law_1.2,0.2917632102966309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,16,power_law_1.2,0.1503424048423767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,16,power_law_1.01,0.08104959726333619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,16,power_law_1.2,0.3023296117782593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,16,power_law_1.2,0.1482367992401123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,16,power_law_1.2,0.3244415998458862
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,16,power_law_1.01,0.08282880187034607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,16,power_law_1.2,0.1576192021369934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,16,power_law_1.01,0.0861952006816864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,16,power_law_1.2,0.38112640380859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,16,power_law_1.2,0.11520639657974244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,16,power_law_1.2,0.4271423816680908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,16,power_law_1.2,0.14326399564743042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,16,power_law_1.2,0.5074624061584473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,16,power_law_1.2,0.1626688003540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,16,power_law_1.2,0.7613503932952881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,16,power_law_1.2,0.6188543796539306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,16,power_law_1.2,0.3048831939697266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,16,power_law_1.2,0.29827840328216554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,16,power_law_1.2,0.7872511863708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,16,power_law_1.2,0.9867712020874023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,16,power_law_1.2,0.30760960578918456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,16,power_law_1.2,0.9591679573059082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,16,power_law_1.2,0.3229824066162109
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,16,power_law_1.01,0.0871999979019165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,16,power_law_1.2,1.1834752082824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,16,power_law_1.2,0.32439680099487306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,16,power_law_1.01,0.0846783995628357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,16,power_law_1.2,0.3510272026062012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,16,power_law_1.2,1.3354559898376466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,16,power_law_1.01,0.09185280203819275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,16,power_law_1.2,1.6493759155273438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,16,power_law_1.2,0.3523200035095215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,16,power_law_1.01,0.08966400027275086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,16,power_law_1.2,0.3933248043060303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,16,power_law_1.01,0.09629439711570739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,16,power_law_1.2,1.8066751480102539
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,16,power_law_1.01,0.11424640417099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,16,power_law_1.2,0.46177921295166013
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,16,power_law_1.01,0.118886399269104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,16,power_law_1.2,0.5432447910308837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,16,power_law_1.01,0.10937600135803223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,16,power_law_1.01,0.20312960147857667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,16,power_law_1.2,2.458176040649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,16,power_law_1.2,0.6169151782989502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,16,power_law_1.01,0.17521920204162597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,16,power_law_1.2,0.8117119789123535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,16,power_law_1.01,0.14020479917526246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,16,power_law_1.01,0.127839994430542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,16,power_law_1.2,2.106515121459961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,16,power_law_1.2,0.9835519790649414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,16,power_law_1.01,0.14394880533218385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,16,power_law_1.2,3.2942081451416017
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,16,power_law_1.01,0.1682752013206482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,16,power_law_1.2,1.333561611175537
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,16,power_law_1.01,0.19743360280990602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,16,power_law_1.01,0.25216639041900635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,16,power_law_1.2,3.205414581298828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,16,power_law_1.2,1.7846656799316407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,16,power_law_1.2,2.5905344009399416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,16,power_law_1.2,4.138175964355469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,16,1,balanced,0.03949866692225138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,16,1,balanced,0.04995200037956238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,16,1,balanced,0.0680159976085027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,16,1,balanced,0.10932266712188721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,16,1,balanced,0.22209066152572632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,16,1,balanced,0.38815999031066895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,16,1,balanced,0.35632534821828205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,16,power_law_1.01,0.31910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,16,1,balanced,0.3574560085932414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,16,1,balanced,0.2976213296254476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,16,1,balanced,0.3004159927368164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,16,power_law_1.01,0.43944320678710935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,16,power_law_1.2,3.7730369567871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,16,1,balanced,0.3760266701380412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,16,1,balanced,0.3733866612116496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,16,1,balanced,0.37751468022664386
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,16,power_law_1.01,0.575648021697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,16,1,balanced,0.38044265906016034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,16,1,balanced,0.3896266619364421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,16,1,balanced,0.395248015721639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,16,1,balanced,0.4070880015691121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,16,power_law_1.01,0.8377535820007325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,16,1,balanced,0.4906933307647705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,16,1,balanced,0.5092639923095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,16,1,balanced,0.5498559872309366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,16,1,balanced,0.5888586839040121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,16,power_law_1.01,1.1843903541564942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,16,power_law_1.2,4.699251174926758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,16,1,balanced,0.6699679692586263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,16,1,balanced,0.03148799886306127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,16,1,balanced,0.03387200087308884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,16,1,balanced,0.04763199885686239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,16,1,balanced,0.0699946681658427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,16,1,balanced,0.7555039723714193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,16,power_law_1.2,0.05675519704818725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,16,1,balanced,0.114464004834493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,16,1,balanced,0.20082134008407593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,16,1,balanced,0.24734399716059366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,16,power_law_1.2,0.08202880024909973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,16,1,balanced,1.1582667032877605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,16,1,balanced,0.2467093269030253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,16,1,balanced,0.22210667530695596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,16,1,balanced,0.2245546579360962
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,16,power_law_1.2,0.07664639949798584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,16,1,balanced,0.2754240036010742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,16,1,balanced,1.3252960046132405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,16,1,balanced,0.27722134192784625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,16,power_law_1.2,0.08177279829978942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,16,1,balanced,0.28226667642593384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,16,1,balanced,0.28842665751775104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,16,1,balanced,0.2942720055580139
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,16,power_law_1.2,0.08227840065956116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,16,1,balanced,1.9192746480305989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,16,1,balanced,0.3020640015602112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,16,1,balanced,0.31305599212646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,16,1,balanced,0.4020693302154541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,16,power_law_1.2,0.08255360126495362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,16,1,balanced,0.42280534903208417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,16,power_law_1.2,6.500057220458984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,16,power_law_1.2,0.08302080035209655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,16,1,balanced,0.4608159859975179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,16,1,balanced,2.46834659576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,16,1,balanced,0.5050133466720581
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,16,power_law_1.2,0.09104639887809754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,16,1,balanced,0.5923893451690674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,16,1,balanced,0.6792106628417969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,16,power_law_1.2,0.0859391987323761
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,16,power_law_1.2,0.08543999791145325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,16,1,balanced,1.044437328974406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,16,power_law_1.2,0.09327999949455261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,16,1,balanced,1.2099786599477131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,16,power_law_1.2,0.11318399906158447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,16,power_law_1.2,0.10432640314102173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,16,1,balanced,1.7466773986816406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,16,power_law_1.2,0.10833920240402221
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,16,power_law_1.2,0.2073983907699585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,16,1,balanced,2.269370714823405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,16,power_law_1.2,0.19924479722976685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,16,1,power_law_1.01,0.03772160112857818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,16,power_law_1.2,0.13381760120391845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,16,1,power_law_1.01,0.04702720046043396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,16,power_law_1.2,0.1417088031768799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,16,1,power_law_1.01,0.06668800115585327
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,16,power_law_1.2,0.1474303960800171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,16,1,power_law_1.01,0.10602240562438965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,16,power_law_1.2,0.1809280037879944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,16,1,power_law_1.01,0.15877120494842528
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,16,power_law_1.2,0.21136000156402587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,16,1,power_law_1.01,0.23264639377593993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,16,power_law_1.2,0.2857408046722412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,16,1,power_law_1.01,0.2953984022140503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,16,1,balanced,0.029359998802344005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,16,1,balanced,0.029509333272775013
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,16,1,balanced,0.033258666594823204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,16,1,balanced,0.04554133117198944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,16,1,power_law_1.01,0.30379519462585447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,16,power_law_1.2,0.35383679866790774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,16,1,balanced,0.0722453345855077
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,16,1,balanced,0.11479467153549194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,16,1,balanced,0.11957333485285442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,16,1,power_law_1.01,0.243174409866333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,16,1,balanced,0.12338667114575703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,16,1,balanced,0.12541332840919495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,16,power_law_1.2,0.49850878715515134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,16,1,balanced,0.12804800271987915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,16,1,power_law_1.01,0.24403200149536133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,16,1,balanced,0.1320319970448812
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,16,1,balanced,0.13784000277519226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,16,1,balanced,0.1440000037352244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,16,1,power_law_1.01,0.3050879955291748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,16,1,balanced,0.15414933363596597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,16,1,balanced,0.18065067132314047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,16,1,balanced,0.18900799751281738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,16,1,power_law_1.01,0.2964735984802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,16,1,balanced,0.20879467328389487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,16,1,balanced,0.21177067359288534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,16,1,power_law_1.01,0.2984256029129028
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,16,1,balanced,0.23837333917617798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,16,1,balanced,0.2630346616109212
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,16,1,balanced,0.3018346627553304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,16,1,power_law_1.01,0.3132352113723755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,16,1,balanced,0.420085350672404
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,16,1,balanced,0.455077330271403
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,16,1,balanced,0.6380319992701212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,16,1,power_law_1.2,1.6162879943847657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,16,1,balanced,0.6708587010701498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,16,1,power_law_1.01,0.03328000009059906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,16,power_law_1.2,0.6080319881439209
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,16,1,balanced,0.958405335744222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,16,1,power_law_1.01,0.037676799297332766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,16,1,balanced,1.243839979171753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,16,1,power_law_1.01,0.05406079888343811
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,16,power_law_1.2,0.916652774810791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,16,1,power_law_1.01,0.07374079823493958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,16,1,power_law_1.01,0.09432960152626038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,16,1,power_law_1.01,0.3374399900436401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,16,1,power_law_1.01,0.1265023946762085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,16,1,power_law_1.01,0.36482560634613037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,16,power_law_1.2,1.3247936248779297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,16,1,power_law_1.01,0.187718403339386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,16,1,power_law_1.2,0.8391488075256348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,16,1,power_law_1.01,0.38888959884643554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,16,1,power_law_1.01,0.19326080083847047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,16,1,power_law_1.01,0.47491841316223143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,16,1,power_law_1.01,0.1887935996055603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,16,1,power_law_1.01,0.19406720399856567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,16,1,power_law_1.01,0.5076352119445801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,16,1,power_law_1.01,0.2285952091217041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,16,1,power_law_1.01,0.5622208118438721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,16,1,power_law_1.01,0.23463680744171142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,16,1,power_law_1.01,0.2395263910293579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,16,1,power_law_1.01,0.26771841049194334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,16,1,power_law_1.01,0.27623040676116944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,16,1,power_law_1.01,0.2854207992553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,16,1,power_law_1.01,0.30404479503631593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,16,1,power_law_1.2,1.9982463836669921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,16,1,power_law_1.01,0.3903808116912842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,16,1,power_law_1.01,0.4211264133453369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,16,1,power_law_1.01,0.6671232223510742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,16,1,power_law_1.01,0.47703042030334475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,16,1,power_law_1.01,0.7844799995422364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,16,1,power_law_1.01,0.5627520084381104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,16,1,power_law_1.01,0.6793087959289551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,16,1,power_law_1.01,0.9009920120239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,16,1,power_law_1.01,0.8010687828063965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,16,1,power_law_1.01,0.029356798529624938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,16,1,power_law_1.01,1.1910592079162599
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,16,1,power_law_1.01,0.03054719865322113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,16,1,power_law_1.01,0.03498240113258362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,16,1,power_law_1.01,1.4755519866943358
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,16,1,power_law_1.01,0.04838399887084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,16,1,power_law_1.01,0.06552960276603699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,16,1,power_law_1.01,2.0532608032226562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,16,1,power_law_1.01,0.08562560081481933
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,16,1,power_law_1.01,0.10470399856567383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,16,1,power_law_1.01,1.071987247467041
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,16,1,power_law_1.01,0.1091647982597351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,16,1,power_law_1.01,2.641926383972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,16,1,power_law_1.01,0.1128000020980835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,16,1,power_law_1.01,1.3319680213928222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,16,1,power_law_1.01,0.1191167950630188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,16,1,power_law_1.01,0.12513279914855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,16,1,power_law_1.01,0.13338240385055541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,16,1,power_law_1.01,1.869068717956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,16,1,power_law_1.01,0.14237439632415771
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,16,1,power_law_1.01,0.157151997089386
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,16,1,power_law_1.01,0.17134079933166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,16,1,power_law_1.01,2.4088640213012695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,16,1,power_law_1.01,0.1858240008354187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,16,1,power_law_1.2,0.03776000142097473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,16,1,power_law_1.2,0.02991360127925873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,16,1,power_law_1.01,0.21522560119628906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,16,1,power_law_1.2,0.0486272007226944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,16,1,power_law_1.2,0.03383679986000061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,16,1,power_law_1.01,0.211897611618042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,16,1,power_law_1.2,0.06626560091972351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,16,1,power_law_1.2,0.0495743989944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,16,1,power_law_1.01,0.24669439792633058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,16,1,power_law_1.2,0.10398720502853394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,16,1,power_law_1.2,0.06746240258216858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,16,1,power_law_1.2,0.4246335983276367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,16,1,power_law_1.01,0.26565120220184324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,16,1,power_law_1.2,0.14485119581222533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,16,1,power_law_1.2,0.08792319893836975
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,16,1,power_law_1.01,0.32003839015960694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,16,1,power_law_1.2,0.2170880079269409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,16,1,power_law_1.2,0.11516159772872925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,16,1,power_law_1.01,0.39292800426483154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,16,1,power_law_1.2,0.28903679847717284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,16,1,power_law_1.2,0.18513920307159423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,16,1,power_law_1.01,0.48613758087158204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,16,1,power_law_1.2,0.1898751974105835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,16,1,power_law_1.2,0.2961983919143677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,16,1,power_law_1.2,0.17970559597015381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,16,1,power_law_1.2,0.23653120994567872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,16,1,power_law_1.2,0.19697920083999634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,16,1,power_law_1.2,0.24730238914489747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,16,1,power_law_1.2,0.2304896116256714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,16,1,power_law_1.2,0.2911423921585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,16,1,power_law_1.2,0.23134078979492187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,16,1,power_law_1.2,0.2988991975784302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,16,1,power_law_1.2,0.24126079082489013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,16,1,power_law_1.2,0.30690560340881345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,16,1,power_law_1.2,0.265068793296814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,16,1,power_law_1.2,0.35212159156799316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,16,1,power_law_1.2,0.2854975938796997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,16,1,power_law_1.2,0.3686336040496826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,16,1,power_law_1.2,0.2920959949493408
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,16,1,power_law_1.01,0.5653376102447509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,16,1,power_law_1.2,0.38007678985595705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,16,1,power_law_1.2,0.3113152027130127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,16,1,power_law_1.01,0.7160128116607666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,16,1,power_law_1.2,0.4037439823150635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,16,1,power_law_1.2,0.42990717887878416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,16,1,power_law_1.01,0.9956992149353028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,16,1,power_law_1.2,0.518342399597168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,16,1,power_law_1.2,0.5753024101257325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,16,1,power_law_1.01,1.3359295845031738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,16,1,power_law_1.2,0.6904960155487061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,16,1,power_law_1.2,0.39888639450073243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,16,1,power_law_1.2,0.836467170715332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,16,1,power_law_1.2,0.48604159355163573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,16,1,power_law_1.2,1.0940352439880372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,16,1,power_law_1.2,0.5187392234802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,16,1,power_law_1.2,0.6115456104278565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,16,1,power_law_1.2,1.350822353363037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,16,1,power_law_1.2,0.6855679988861084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,16,1,power_law_1.2,0.7978240013122558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,16,1,power_law_1.2,1.8843391418457032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,16,1,power_law_1.2,0.9559871673583984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,16,1,power_law_1.2,2.429030418395996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,16,1,power_law_1.2,1.2155520439147949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,16,1,power_law_1.2,1.496448040008545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,16,1,power_law_1.2,2.080371284484863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,16,1,power_law_1.2,2.6654336929321287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,16,balanced,0.04390933116277059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,16,balanced,0.04372266431649526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,16,balanced,0.058090666929880776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,16,balanced,0.07860800127188365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,16,balanced,0.1406613290309906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,16,balanced,0.23507734139760336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,16,balanced,0.23222933212916055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,16,balanced,0.23244800170262656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,16,balanced,0.2485546668370565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,16,balanced,0.2508693337440491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,16,balanced,0.3116640051205953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,16,balanced,0.31566399335861206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,16,balanced,0.32118932406107586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,16,balanced,0.3293493390083313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,16,balanced,0.33691199620564777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,16,balanced,0.3481866518656413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,16,balanced,0.3677226702372233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,16,balanced,0.4169546763102214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,16,1,power_law_1.2,0.026931199431419372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,16,balanced,0.4448586702346802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,16,balanced,0.5011946757634481
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,16,1,power_law_1.2,0.029260799288749695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,16,balanced,0.5567200183868408
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,16,1,power_law_1.2,0.03406080007553101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,16,balanced,0.6695626576741537
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,16,1,power_law_1.2,0.044614401459693906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,16,balanced,0.7849120299021403
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,16,1,power_law_1.2,0.06156799793243408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,16,balanced,1.130405346552531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,16,1,power_law_1.2,0.08283519744873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,16,balanced,1.3617653846740723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,16,1,power_law_1.2,0.1015936017036438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,16,1,power_law_1.2,0.10788480043411255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,16,balanced,1.953770637512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,16,1,power_law_1.2,0.1121216058731079
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,16,1,power_law_1.2,0.11735039949417114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,16,balanced,2.541306654612223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,16,1,power_law_1.2,0.12392959594726563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,16,1,power_law_1.2,0.1332479953765869
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,16,1,power_law_1.2,0.14120960235595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,16,1,power_law_1.2,0.15693440437316894
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,16,1,power_law_1.2,0.1726464033126831
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,16,1,power_law_1.2,0.187391996383667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,16,1,power_law_1.2,0.21591041088104249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,16,1,power_law_1.2,0.21415679454803466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,16,1,power_law_1.2,0.2509119987487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,16,1,power_law_1.2,0.2692608118057251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,16,1,power_law_1.2,0.32242560386657715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,16,1,power_law_1.2,0.3982784032821655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,16,1,power_law_1.2,0.4874879837036133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,16,1,power_law_1.2,0.5739071846008301
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,16,1,power_law_1.2,0.7209407806396484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,16,1,power_law_1.2,1.0065919876098632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,16,balanced,0.037658666570981346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,16,balanced,0.03790933390458425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,16,balanced,0.04386133452256521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,16,balanced,0.06432533264160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,16,balanced,0.0869706670443217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,16,balanced,0.1460640033086141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,16,balanced,0.1689173380533854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,16,1,power_law_1.2,1.3576000213623047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,16,balanced,0.17108800013860068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,16,balanced,0.1585973302523295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,16,balanced,0.16285333037376404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,16,balanced,0.2312426765759786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,16,balanced,0.24156266450881958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,16,balanced,0.24710933367411295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,16,balanced,0.25701866547266644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,16,balanced,0.26708799600601196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,16,balanced,0.27568533023198444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,16,balanced,0.2974399924278259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,16,balanced,0.3349386850992839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,16,balanced,0.36526934305826825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,16,balanced,0.42790400981903076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,16,balanced,0.49033598105112713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,16,balanced,0.6147520144780477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,16,balanced,0.739578644434611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,16,balanced,0.05407466491063436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,16,balanced,0.05605333546797434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,16,balanced,0.06653866668542226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,16,balanced,1.0505173206329346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,16,balanced,0.08126933375994365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,16,balanced,0.0990559955437978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,16,balanced,0.14696533481280008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,16,balanced,0.1492853363355001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,16,balanced,1.2998560269673665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,16,balanced,1.8639893531799316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,16,balanced,2.419930617014567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,16,balanced,0.15268266201019287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,16,balanced,0.28734399875005084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,16,balanced,0.28948267300923664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,16,balanced,0.2967573404312134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,16,balanced,0.3036266764005025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,16,balanced,0.3087093234062195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,16,balanced,0.32441065708796185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,16,balanced,0.33674665292104083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,16,balanced,0.3508479992548625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,16,balanced,0.37518401940663654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,16,balanced,0.42206935087839764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,16,balanced,0.47177600860595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,16,balanced,0.5645386775334676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,16,balanced,0.6585866610209147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,16,balanced,0.9707732995351156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,16,balanced,1.1571306387583415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,16,balanced,1.6584107081095378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,16,balanced,2.154693285624186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,16,balanced,0.029525332152843475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,16,balanced,0.0315733328461647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,16,balanced,0.03148266673088074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,16,balanced,3.1469974517822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,16,balanced,0.036357333262761436
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,16,balanced,0.05820266902446747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,16,balanced,0.08270933230717976
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,16,balanced,0.08301866551240285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,16,balanced,0.08710400263468425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,16,balanced,0.08717866738637288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,16,balanced,0.08749333024024963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,16,balanced,0.08877866466840108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,16,balanced,0.12268799543380737
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,16,balanced,0.12651733557383218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,16,balanced,0.11455999811490376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,16,balanced,4.1375306447347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,16,balanced,0.2749279936154683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,16,balanced,0.22817599773406982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,16,power_law_1.01,0.13100800514221192
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,16,balanced,0.16694400707880655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,16,balanced,0.12265066305796306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,16,balanced,0.12946133812268576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,16,power_law_1.01,0.21671679019927978
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,16,balanced,0.14517866571744284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,16,balanced,0.16265599926312765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,16,power_law_1.01,0.21544320583343507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,16,balanced,0.19875200589497885
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,16,balanced,0.23693867524464926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,16,power_law_1.01,0.20280320644378663
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,16,balanced,0.3359786669413249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,16,power_law_1.01,0.21982719898223876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,16,balanced,0.4102026621500651
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,16,balanced,0.583407998085022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,16,power_law_1.01,0.19773440361022948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,16,balanced,0.7606613636016846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,16,power_law_1.01,0.22029440402984618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,16,power_law_1.01,0.22190721035003663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,16,power_law_1.01,0.07858560085296631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,16,power_law_1.01,0.2292095899581909
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,16,power_law_1.01,0.13883520364761354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,16,power_law_1.01,0.2248255968093872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,16,power_law_1.01,0.13843200206756592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,16,power_law_1.01,0.27948799133300783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,16,power_law_1.01,0.13564159870147705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,16,power_law_1.01,0.29003520011901857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,16,power_law_1.01,0.12934399843215943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,16,power_law_1.01,0.303654408454895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,16,power_law_1.01,0.1321023941040039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,16,power_law_1.01,0.31640961170196535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,16,power_law_1.01,0.1449280023574829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,16,power_law_1.01,0.3249408006668091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,16,power_law_1.01,0.15798399448394776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,16,power_law_1.01,0.3214783906936646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,16,power_law_1.01,0.1563199996948242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,16,power_law_1.01,0.3590399980545044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,16,power_law_1.01,0.14606720209121704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,16,power_law_1.01,0.4088448047637939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,16,power_law_1.01,0.21621758937835694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,16,power_law_1.01,0.45905280113220215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,16,power_law_1.01,0.23212800025939942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,16,power_law_1.01,0.5418432235717774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,16,power_law_1.01,0.23014400005340577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,16,power_law_1.01,0.24724481105804444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,16,power_law_1.01,0.6121920108795166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,16,power_law_1.01,0.26112639904022217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,16,power_law_1.01,0.8048576354980469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,16,power_law_1.01,0.2680959939956665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,16,power_law_1.01,0.29525759220123293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,16,power_law_1.01,0.9532992362976074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,16,power_law_1.01,0.33408639430999754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,16,power_law_1.01,0.3729471921920776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,16,power_law_1.01,1.3284480094909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,16,power_law_1.01,0.45316481590270996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,16,power_law_1.01,0.5256576061248779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,16,power_law_1.01,1.7430208206176758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,16,power_law_1.01,0.6986815929412842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,16,power_law_1.01,0.8381504058837891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,16,power_law_1.01,2.3210304260253904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,16,power_law_1.01,1.152575969696045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,16,power_law_1.01,1.4554368019104005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,16,power_law_1.01,3.0932416915893555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,16,power_law_1.01,0.08983039855957031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,16,power_law_1.01,0.13822720050811768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,16,power_law_1.01,2.126918411254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,16,power_law_1.01,0.1403264045715332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,16,power_law_1.01,0.13928960561752318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,16,1,power_law_1.2,1.1043264389038085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,16,power_law_1.01,0.14236799478530884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,16,power_law_1.01,2.7246719360351563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,16,power_law_1.01,0.12594560384750367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,16,power_law_1.2,0.13352960348129272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,16,power_law_1.01,0.14176640510559083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,16,power_law_1.2,0.2143359899520874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,16,power_law_1.01,0.1367616057395935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,16,power_law_1.2,0.21262080669403077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,16,power_law_1.01,0.2629120111465454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,16,power_law_1.2,0.21265919208526612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,16,power_law_1.01,0.2681600093841553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,16,1,balanced,0.04786133269468943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,16,power_law_1.2,0.22057600021362306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,16,power_law_1.01,0.2768064022064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,16,power_law_1.2,0.19071359634399415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,16,power_law_1.2,0.21227519512176513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,16,power_law_1.2,0.22382080554962158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,16,power_law_1.2,0.21337599754333497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,16,power_law_1.2,0.22574079036712646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,16,power_law_1.2,0.2953279972076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,16,power_law_1.2,0.2832319974899292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,16,power_law_1.01,0.29614078998565674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,16,power_law_1.2,0.3002624034881592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,16,power_law_1.01,0.295199990272522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,16,power_law_1.2,0.3179712057113647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,16,power_law_1.01,0.3022016048431396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,16,power_law_1.01,0.05198079943656921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,16,power_law_1.01,0.32058238983154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,16,power_law_1.01,0.3359551906585693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,16,power_law_1.01,0.07626879811286927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,16,power_law_1.01,0.39867520332336426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,16,power_law_1.01,0.0754368007183075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,16,power_law_1.01,0.4652991771697998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,16,power_law_1.01,0.0755840003490448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,16,power_law_1.01,0.07879679799079894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,16,power_law_1.01,0.5340415954589843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,16,power_law_1.01,0.07486079931259156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,16,power_law_1.01,0.6780543804168702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,16,power_law_1.2,0.3346368074417114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,16,1,balanced,0.052149335543314614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,16,1,balanced,0.06800533334414165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,16,1,balanced,0.08673066894213359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,16,power_law_1.2,0.3376768112182617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,16,power_law_1.01,0.8429311752319336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,16,power_law_1.2,0.3601088047027588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,16,power_law_1.2,0.4315328121185303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,16,power_law_1.01,1.1865535736083985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,16,power_law_1.2,0.47823362350463866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,16,power_law_1.01,1.4555583953857423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,16,power_law_1.2,0.560588788986206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,16,power_law_1.01,0.07824000120162963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,16,power_law_1.2,0.6448895931243896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,16,power_law_1.01,0.08007680177688599
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,16,1,power_law_1.2,0.6081535816192627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,16,power_law_1.01,2.052947235107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,16,power_law_1.01,0.0813759982585907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,16,power_law_1.2,0.852627182006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,16,power_law_1.01,0.08358399868011475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,16,power_law_1.01,0.08201599717140198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,16,power_law_1.2,1.0634112358093262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,16,power_law_1.01,0.10906239748001098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,16,power_law_1.01,2.66046085357666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,16,power_law_1.01,0.10792319774627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,16,power_law_1.2,1.4077119827270508
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,16,power_law_1.01,0.0985152006149292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,16,power_law_1.01,0.22254080772399903
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,16,power_law_1.01,0.14968960285186766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,16,power_law_1.2,1.8764671325683593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,16,power_law_1.01,0.13050240278244019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,16,power_law_1.01,3.942617416381836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,16,power_law_1.01,0.11636480093002319
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,16,power_law_1.01,0.1316480040550232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,16,power_law_1.2,2.7733951568603517
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,16,power_law_1.01,0.15167360305786132
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,16,power_law_1.01,0.17313920259475707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,16,power_law_1.01,0.22259199619293213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,16,power_law_1.01,5.115481567382813
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,16,power_law_1.01,0.27513599395751953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,16,power_law_1.2,3.461715316772461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,16,power_law_1.2,0.07876480221748353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,16,power_law_1.01,0.38938241004943847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,16,power_law_1.2,0.13496320247650145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,16,power_law_1.2,0.08991360068321227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,16,power_law_1.01,0.496127986907959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,16,power_law_1.2,0.12090239524841309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,16,power_law_1.2,0.13872640132904052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,16,power_law_1.2,0.13628159761428832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,16,power_law_1.01,0.7397247791290283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,16,power_law_1.2,0.12251520156860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,16,power_law_1.2,0.1391808032989502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,16,1,balanced,0.1269439955552419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,16,power_law_1.2,0.14008320569992067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,16,power_law_1.2,0.129094398021698
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,16,power_law_1.01,1.0643775939941407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,16,power_law_1.2,0.14240000247955323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,16,power_law_1.2,0.14733439683914185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,16,power_law_1.2,0.10741759538650512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,16,power_law_1.2,0.15233919620513917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,16,power_law_1.2,0.12855679988861085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,16,power_law_1.2,0.15523200035095214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,16,power_law_1.2,0.1542847990989685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,16,power_law_1.2,0.13462400436401367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,16,power_law_1.2,0.2160576105117798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,16,power_law_1.2,0.26385281085968015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,16,power_law_1.2,0.2342655897140503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,16,power_law_1.2,0.2608256101608276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,16,1,power_law_1.01,0.044870400428771974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,16,power_law_1.2,0.23832321166992188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,16,power_law_1.2,0.2711679935455322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,16,power_law_1.2,0.25126399993896487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,16,power_law_1.2,0.2799743890762329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,16,power_law_1.2,0.26175360679626464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,16,power_law_1.2,0.2987776041030884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,16,power_law_1.2,0.2713279962539673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,16,1,power_law_1.2,2.6260095596313477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,16,power_law_1.2,0.29991679191589354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,16,power_law_1.2,0.28959360122680666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,16,1,balanced,0.0352906659245491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,16,1,balanced,0.04553600152333578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,16,1,balanced,0.06005333364009857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,16,power_law_1.2,0.3395071983337402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,16,1,balanced,0.09564266602198283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,16,power_law_1.2,0.34603519439697267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,16,1,balanced,0.19222400585810342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,16,1,balanced,0.33538134892781574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,16,power_law_1.2,0.3550656080245972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,16,1,balanced,0.3077813386917114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,16,power_law_1.2,0.37674241065979003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,16,1,balanced,0.30875200033187866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,16,1,balanced,0.2563733259836833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,16,power_law_1.2,0.39891839027404785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,16,1,balanced,0.2568693359692891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,16,power_law_1.2,0.4653183937072754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,16,1,balanced,0.3238133390744527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,16,1,balanced,0.3221279978752136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,16,power_law_1.2,0.4708415985107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,16,1,balanced,0.3248479962348938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,16,power_law_1.2,0.5463295936584472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,16,1,balanced,0.3287573258082072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,16,1,balanced,0.3344053427378337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,16,power_law_1.2,0.553004789352417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,16,1,balanced,0.34010668595631915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,16,power_law_1.2,0.7125376224517822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,16,1,balanced,0.34902934233347577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,16,power_law_1.2,0.0521664023399353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,16,1,balanced,0.4257599910100301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,16,power_law_1.2,0.7296576023101806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,16,1,balanced,0.4410773515701294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,16,power_law_1.2,0.8749183654785156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,16,power_law_1.2,0.07511039972305297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,16,1,balanced,0.47572799523671466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,16,1,balanced,0.5098613500595093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,16,power_law_1.2,0.07133439779281617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,16,power_law_1.2,0.8871552467346191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,16,1,balanced,0.5798879861831665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,16,power_law_1.2,1.232639980316162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,16,power_law_1.2,0.0753600001335144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,16,1,balanced,0.6533066829045614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,16,power_law_1.2,1.165401554107666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,16,1,balanced,1.0036319891611736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,16,power_law_1.2,1.5680895805358888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,16,1,balanced,1.1488640308380127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,16,1,balanced,0.20484799146652222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,16,1,balanced,0.20736533403396606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,16,1,balanced,0.20732800165812174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,16,power_law_1.2,1.575321578979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,16,1,power_law_1.01,0.05007359981536865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,16,1,balanced,1.6699306170145671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,16,1,power_law_1.01,0.0680895984172821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,16,power_law_1.2,2.2166080474853516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,16,1,power_law_1.01,0.08360959887504578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,16,1,balanced,2.1423999468485513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,16,power_law_1.2,2.1632320404052736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,16,1,balanced,0.02959466725587845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,16,1,balanced,0.03357866654793421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,16,1,balanced,0.04599999884764353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,16,1,balanced,0.063701331615448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,16,power_law_1.2,0.07834879755973816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,16,power_law_1.2,0.07426559925079346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,16,power_law_1.2,2.8660863876342773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,16,power_law_1.2,0.07645440101623535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,16,power_law_1.2,2.9265024185180666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,16,power_law_1.2,0.07909759879112244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,16,power_law_1.2,0.07505279779434204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,16,power_law_1.2,0.08332800269126892
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,16,power_law_1.2,0.08287360072135926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,16,1,balanced,0.10054399569829305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,16,1,balanced,0.18156800667444864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,16,1,balanced,0.21486934026082358
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,16,power_law_1.2,0.09969279766082764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,16,1,balanced,0.2143146594365438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,16,power_law_1.2,4.217932891845703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,16,1,balanced,0.1946559945742289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,16,1,balanced,0.19408533970514932
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,16,power_law_1.2,0.10390399694442749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,16,1,balanced,0.24075732628504434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,16,1,balanced,0.24285866816838583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,16,power_law_1.2,0.10080000162124633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,16,1,balanced,0.24532800912857056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,16,1,balanced,0.25171732902526855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,16,power_law_1.2,0.20784640312194824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,16,1,balanced,0.25515200694402057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,16,1,balanced,0.2619253396987915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,16,1,balanced,0.27162132660547894
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,16,power_law_1.2,0.16194560527801513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,16,1,balanced,0.3513706525166829
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,16,power_law_1.2,0.12936320304870605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,16,power_law_1.2,0.12212480306625366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,16,power_law_1.2,0.13569920063018798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,16,power_law_1.2,5.824665451049805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,16,power_law_1.2,0.15641599893569946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,16,power_law_1.2,0.18570879697799683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,16,power_law_1.2,0.25589759349823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,16,1,balanced,0.36894933382670086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,16,1,balanced,0.40381332238515216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,16,1,balanced,0.44196800390879315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,16,1,balanced,0.5169066588083903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,16,1,balanced,0.5957440137863159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,16,1,balanced,0.912933349609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,16,1,balanced,1.0622933705647786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,16,1,balanced,1.5347092946370442
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,16,power_law_1.2,0.32303359508514407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,16,1,balanced,0.3890293439229329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,16,power_law_1.2,0.4501952171325684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,16,1,balanced,0.39260268211364746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,16,1,balanced,0.3949173291524251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,16,1,balanced,0.39954666296641034
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,16,power_law_1.2,0.5587711811065674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,16,1,balanced,0.4046666622161865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,16,1,balanced,0.41125333309173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,16,1,balanced,0.41911999384562176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,16,power_law_1.2,0.782854413986206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,16,1,power_law_1.2,0.04635519981384277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,16,1,balanced,0.4267093340555827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,16,1,balanced,0.4383680025736491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,16,1,balanced,0.46593066056569415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,16,1,balanced,0.492576003074646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,16,1,balanced,0.5444693168004354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,16,1,power_law_1.01,0.10898560285568237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,16,1,balanced,1.985978603363037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,16,1,balanced,0.6009813149770101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,16,1,power_law_1.01,0.033548799157142636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,16,1,balanced,0.9616906642913818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,16,1,power_law_1.01,0.040191999077796935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,16,1,balanced,1.069706678390503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,16,1,balanced,0.028725333511829376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,16,1,power_law_1.01,0.05857920050621033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,16,1,balanced,0.03012266755104065
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,16,1,balanced,0.03311999887228012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,16,1,balanced,0.04188266893227895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,16,1,power_law_1.01,0.09377920031547546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,16,1,balanced,0.06653333206971486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,16,1,balanced,1.5407147407531738
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,16,power_law_1.2,1.1987775802612304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,16,1,balanced,0.10299733281135559
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,16,1,balanced,0.10520533720652263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,16,1,balanced,0.11001066366831462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,16,1,power_law_1.01,0.1380288004875183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,16,1,power_law_1.01,0.028908801078796387
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,16,1,balanced,0.11237866679827373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,16,1,balanced,0.11395200093587239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,16,1,balanced,1.9577280680338542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,16,1,power_law_1.01,0.2057728052139282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,16,1,power_law_1.01,0.03131519854068756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,16,1,power_law_1.01,0.260096001625061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,16,1,power_law_1.01,0.0441536009311676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,16,1,power_law_1.01,0.06241919994354248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,16,1,power_law_1.01,0.2581248044967651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,16,1,balanced,2.885471979777018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,16,1,power_law_1.01,0.08638719916343689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,16,1,power_law_1.01,0.20981121063232422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,16,1,power_law_1.01,0.21191680431365967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,16,1,power_law_1.01,0.10936319828033447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,16,1,power_law_1.01,0.25309441089630125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,16,1,power_law_1.01,0.16463359594345092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,16,1,balanced,3.77403195699056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,16,1,power_law_1.01,0.25649280548095704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,16,1,power_law_1.01,0.1683519959449768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,16,1,balanced,0.11647466818491618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,16,1,balanced,0.12152533729871114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,16,1,balanced,0.12487999598185222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,16,1,power_law_1.01,0.2700479984283447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,16,1,power_law_1.01,0.160697603225708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,16,1,balanced,0.1349013348420461
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,16,1,balanced,0.15875732898712158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,16,1,power_law_1.01,0.2755903959274292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,16,1,balanced,0.16671466827392578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,16,1,power_law_1.01,0.16958719491958618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,16,1,power_law_1.2,0.05235840082168579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,16,1,power_law_1.01,0.3003135919570923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,16,1,power_law_1.01,0.20156800746917725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,16,1,power_law_1.2,0.06863999962806702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,16,1,power_law_1.01,0.3233664035797119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,16,1,power_law_1.01,0.2048703908920288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,16,1,power_law_1.2,0.0828544020652771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,16,1,power_law_1.01,0.20887680053710939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,16,1,power_law_1.01,0.3378943920135498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,16,1,power_law_1.01,0.13423999547958373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,16,1,power_law_1.01,0.22442879676818847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,16,1,power_law_1.01,0.40760321617126466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,16,1,power_law_1.01,0.16019200086593627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,16,1,power_law_1.01,0.23633921146392822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,16,1,power_law_1.01,0.4378176212310791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,16,1,power_law_1.01,0.1696895956993103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,16,1,power_law_1.01,0.25191679000854494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,16,1,power_law_1.01,0.4872896194458008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,16,1,power_law_1.01,0.2670720100402832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,16,1,balanced,0.18339733282725015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,16,1,balanced,0.18657066424687704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,16,1,power_law_1.01,0.5811456203460693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,16,1,power_law_1.01,0.33919999599456785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,16,1,balanced,0.2101759910583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,16,1,balanced,0.23005332549413046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,16,1,power_law_1.01,0.36805119514465334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,16,1,balanced,0.263973335425059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,16,1,power_law_1.01,0.6805503845214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,16,1,balanced,0.3452906608581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,16,1,power_law_1.01,0.41933441162109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,16,1,balanced,0.3972800175348918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,16,1,power_law_1.01,0.7963967800140381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,16,1,balanced,0.5623946587244669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,16,1,power_law_1.01,0.49295997619628906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,16,1,balanced,0.5881760120391846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,16,1,power_law_1.01,0.593075180053711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,16,1,power_law_1.2,0.7985151767730713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,16,1,balanced,0.829637368520101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,16,1,power_law_1.01,0.7049215793609619
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,16,1,balanced,1.0708159605662029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,16,1,power_law_1.2,0.033164799213409424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,16,1,power_law_1.01,0.9402815818786621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,16,1,power_law_1.2,0.04195840060710907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,16,1,power_law_1.2,0.05805439949035644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,16,1,power_law_1.01,1.1643263816833496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,16,1,power_law_1.2,0.09185280203819275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,16,1,power_law_1.01,1.0396032333374023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,16,1,power_law_1.2,0.12646399736404418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,16,1,power_law_1.01,0.041868799924850465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,16,1,power_law_1.01,1.635968017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,16,1,power_law_1.2,0.18876800537109376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,16,1,power_law_1.01,1.2827199935913085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,16,1,power_law_1.2,0.24809598922729492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,16,1,power_law_1.2,0.25910398960113523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,16,1,power_law_1.01,2.1066112518310547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,16,1,power_law_1.01,1.7839168548583983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,16,1,power_law_1.2,0.20238080024719238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,16,1,power_law_1.2,0.028960001468658448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,16,1,power_law_1.2,0.21000320911407472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,16,1,power_law_1.2,0.032204800844192506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,16,1,power_law_1.2,0.2523904085159302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,16,1,power_law_1.2,0.04549759924411774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,16,1,power_law_1.2,0.2571007966995239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,16,1,power_law_1.01,2.290336036682129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,16,1,power_law_1.2,0.061414402723312375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,16,1,power_law_1.2,0.25859200954437256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,16,1,power_law_1.01,0.027923199534416198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,16,1,power_law_1.2,0.08001919984817504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,16,1,power_law_1.2,0.274617600440979
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,16,1,power_law_1.01,0.02974080145359039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,16,1,power_law_1.2,0.10191359519958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,16,1,power_law_1.2,0.32133119106292723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,16,1,power_law_1.01,0.03309440016746521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,16,1,power_law_1.2,0.1627135992050171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,16,1,power_law_1.2,0.32767999172210693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,16,1,power_law_1.2,0.10204800367355346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,16,1,power_law_1.01,0.04376960098743439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,16,1,power_law_1.2,0.16682239770889282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,16,1,power_law_1.2,0.3399168014526367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,16,1,power_law_1.2,0.15728000402450562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,16,1,power_law_1.01,0.05924479961395264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,16,1,power_law_1.2,0.41475839614868165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,16,1,power_law_1.2,0.1599679946899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,16,1,power_law_1.01,0.07899519801139832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,16,1,power_law_1.2,0.4440959930419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,16,1,power_law_1.2,0.2024768114089966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,16,1,power_law_1.01,0.09219200015068055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,16,1,power_law_1.2,0.20723841190338135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,16,1,power_law_1.2,0.5316544055938721
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,16,1,power_law_1.01,0.09910399913787842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,16,1,power_law_1.2,0.21756160259246826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,16,1,power_law_1.2,0.5923840045928955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,16,1,power_law_1.01,0.10110080242156982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,16,1,power_law_1.01,0.046387198567390445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,16,1,power_law_1.01,0.10605440139770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,16,1,power_law_1.01,0.35192320346832273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,16,1,power_law_1.01,0.060108798742294314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,16,1,power_law_1.01,0.3538304090499878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,16,1,power_law_1.01,0.07529600262641907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,16,1,power_law_1.01,0.3634943962097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,16,1,power_law_1.01,0.3697920083999634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,16,1,power_law_1.01,0.3744256019592285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,16,1,power_law_1.2,0.2244352102279663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,16,1,power_law_1.01,0.3835263967514038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,16,1,power_law_1.2,0.24620161056518555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,16,1,power_law_1.2,0.6886847972869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,16,1,power_law_1.01,0.39292159080505373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,16,1,power_law_1.2,0.2548863887786865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,16,1,power_law_1.01,0.111353600025177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,16,1,power_law_1.2,0.26867198944091797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,16,1,power_law_1.01,0.4062464237213135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,16,1,power_law_1.2,0.8291520118713379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,16,1,power_law_1.01,0.11694079637527466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,16,1,power_law_1.2,0.349401593208313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,16,1,power_law_1.01,0.42415361404418944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,16,1,power_law_1.01,0.1248703956604004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,16,1,power_law_1.2,1.0572287559509277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,16,1,power_law_1.2,0.37597439289093015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,16,1,power_law_1.01,0.13949439525604249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,16,1,power_law_1.2,0.13015040159225463
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,16,1,power_law_1.01,0.15185920000076295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,16,1,power_law_1.2,0.44434561729431155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,16,1,power_law_1.2,0.1589568018913269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,16,1,power_law_1.2,1.2968832015991212
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,16,1,power_law_1.01,0.16236159801483155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,16,1,power_law_1.2,0.5041152000427246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,16,1,power_law_1.2,0.1667072057723999
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,16,1,power_law_1.01,0.1894719958305359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,16,1,power_law_1.01,0.1870144009590149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,16,1,power_law_1.2,0.6038144111633301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,16,1,power_law_1.2,1.8085504531860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,16,1,power_law_1.01,0.21653120517730712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,16,1,power_law_1.2,0.7310272216796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,16,1,power_law_1.01,0.23162240982055665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,16,1,power_law_1.01,0.4728640079498291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,16,1,power_law_1.01,0.2773695945739746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,16,1,power_law_1.2,0.9587264060974121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,16,1,power_law_1.2,2.317203140258789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,16,1,power_law_1.01,0.3434240102767944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,16,1,power_law_1.2,1.1832768440246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,16,1,power_law_1.01,0.4203968048095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,16,1,power_law_1.01,0.09564800262451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,16,1,power_law_1.01,0.49375357627868655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,16,1,power_law_1.2,1.6513664245605468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,16,1,power_law_1.01,0.6214079856872559
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,16,1,power_law_1.01,0.8605567932128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,16,1,power_law_1.2,2.1309440612792967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,16,balanced,0.029178666571776073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,16,balanced,0.029114666084448498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,16,balanced,0.0295413335164388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,16,balanced,0.030159999926884968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,16,balanced,0.03536533315976461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,16,balanced,0.03771200031042099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,16,1,power_law_1.01,1.1493375778198243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,16,balanced,0.02940800040960312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,16,balanced,0.02945599953333537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,16,balanced,0.027509334186712902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,16,balanced,0.029557332396507263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,16,balanced,0.03350933392842611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,16,balanced,0.03559466699759165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,16,balanced,0.03965333352486292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,16,balanced,0.039781334499518074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,16,balanced,0.04194133480389913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,16,balanced,0.04167466859022776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,16,balanced,0.046575998266537987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,16,balanced,0.050106664498647056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,16,balanced,0.05130666494369507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,16,balanced,0.05607999861240387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,16,balanced,0.05819199979305267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,16,balanced,0.06227200229962667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,16,balanced,0.07047999898592631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,16,balanced,0.08321600159009297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,16,balanced,0.09915199875831604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,16,balanced,0.12574932972590128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,16,balanced,0.15127999583880106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,16,balanced,0.043578664461771645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,16,balanced,0.21706666549046835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,16,1,power_law_1.2,0.027161601185798644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,16,balanced,0.2693173289299011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,16,balanced,0.3797279993693034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,16,balanced,0.4867306550343831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,16,balanced,0.7039840221405029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,16,balanced,0.9238879680633545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,16,1,power_law_1.01,0.571340799331665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,16,balanced,0.037717332442601524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,16,balanced,0.03695466617743174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,16,balanced,0.037733333806196846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,16,balanced,0.03772799919048945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,16,balanced,0.03979199876387914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,16,1,power_law_1.01,0.1177024006843567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,16,1,power_law_1.01,0.6569471836090088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,16,balanced,0.04229333500067393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,16,1,power_law_1.01,0.14504319429397583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,16,balanced,0.0454773356517156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,16,balanced,0.042917331059773765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,16,1,power_law_1.01,0.7457727909088134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,16,balanced,0.04996799925963084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,16,balanced,0.05373333394527435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,16,1,power_law_1.01,0.1488703966140747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,16,balanced,0.054458667834599815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,16,balanced,0.05612266560395559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,16,balanced,0.06198933223883311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,16,balanced,0.06632000207901001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,16,1,power_law_1.01,0.978758430480957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,16,1,power_law_1.2,0.346560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,16,1,power_law_1.01,1.2035903930664062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,16,1,power_law_1.2,0.3523519992828369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,16,balanced,0.043866669138272606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,16,balanced,0.04390400151411692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,16,balanced,0.04458666841189066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,16,balanced,0.056757330894470215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,16,1,power_law_1.2,0.3623552083969116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,16,balanced,0.05798399945100149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,16,1,power_law_1.01,1.6605567932128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,16,balanced,0.05842133363087972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,16,balanced,0.061477333307266235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,16,balanced,0.06381866832574208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,16,1,power_law_1.2,0.36828160285949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,16,balanced,0.06647466619809468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,16,balanced,0.07144533097743988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,16,balanced,0.07688533266385396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,16,1,power_law_1.2,0.37574400901794436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,16,balanced,0.07600533465544383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,16,balanced,0.08689066767692566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,16,balanced,0.09697600205739339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,16,1,power_law_1.01,2.1162303924560546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,16,balanced,0.11998933553695679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,16,1,power_law_1.2,0.38506240844726564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,16,balanced,0.14281599720319113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,16,balanced,0.20172266165415445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,16,1,power_law_1.2,0.3993664026260376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,16,balanced,0.24621333678563437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,16,balanced,0.3486773173014323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,16,1,power_law_1.2,0.41504640579223634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,16,1,power_law_1.2,0.4296000003814697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,16,1,power_law_1.01,2.973676872253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,16,1,power_law_1.2,0.028486400842666626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,16,1,power_law_1.2,0.03281280100345611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,16,balanced,0.08712533116340637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,16,balanced,0.10523200035095215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,16,balanced,0.12541866302490234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,16,1,power_law_1.2,0.04067200124263763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,16,balanced,0.1673439939816793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,16,balanced,0.20371200640996298
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,16,1,power_law_1.2,0.0572160005569458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,16,1,power_law_1.01,3.8276287078857423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,16,balanced,0.4485653241475423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,16,1,power_law_1.2,0.07560319900512695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,16,balanced,0.6475679874420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,16,balanced,0.8456266721089681
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,16,1,power_law_1.2,0.09037439823150635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,16,1,power_law_1.2,0.48076801300048827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,16,power_law_1.01,0.03315199911594391
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,16,1,power_law_1.2,0.09697279930114747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,16,power_law_1.01,0.033228799700737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,16,1,power_law_1.2,0.5849599838256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,16,power_law_1.01,0.03455359935760498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,16,power_law_1.01,0.03405439853668213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,16,1,power_law_1.2,0.6685503959655762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,16,power_law_1.01,0.03438720107078552
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,16,balanced,0.023183998962243397
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,16,balanced,0.023376000424226124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,16,power_law_1.01,0.037049600481987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,16,balanced,0.29209067424138385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,16,balanced,0.023354666928450268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,16,1,power_law_1.2,0.7982783794403077
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,16,balanced,0.02510933329661687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,16,power_law_1.01,0.04021120071411133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,16,balanced,0.02716800073782603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,16,balanced,0.3710346619288127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,16,balanced,0.029194665451844532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,16,balanced,0.03152533372243246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,16,power_law_1.01,0.0409855991601944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,16,balanced,0.5334879954655966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,16,balanced,0.03257599969704946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,16,balanced,0.033615998923778534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,16,power_law_1.01,0.04433279931545257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,16,balanced,0.03358400116364161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,16,1,power_law_1.2,0.9943360328674317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,16,balanced,0.6915573279062907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,16,power_law_1.01,0.04425599873065948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,16,power_law_1.01,0.050419199466705325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,16,balanced,1.0112373034159343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,16,power_law_1.01,0.05097600221633911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,16,1,power_law_1.2,1.2180992126464845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,16,1,power_law_1.2,0.10035840272903443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,16,power_law_1.01,0.054745602607727054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,16,balanced,1.3348746299743652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,16,power_law_1.01,0.05730559825897217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,16,1,power_law_1.2,0.10443520545959473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,16,power_law_1.01,0.030636799335479737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,16,power_law_1.01,0.06227200031280518
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,16,1,power_law_1.2,0.10938880443572999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,16,1,power_law_1.2,1.6670656204223633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,16,power_law_1.01,0.06889600157737732
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,16,1,power_law_1.2,0.11898239850997924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,16,power_law_1.01,0.07684479951858521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,16,1,power_law_1.01,0.3033087968826294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,16,power_law_1.01,0.08949120044708252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,16,1,power_law_1.2,0.12455040216445923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,16,1,power_law_1.01,0.30697600841522216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,16,power_law_1.01,0.10170880556106568
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,16,balanced,0.03335466732581457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,16,balanced,0.0545653353134791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,16,1,power_law_1.2,0.1382591962814331
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,16,balanced,0.05410666763782501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,16,1,power_law_1.01,0.31515519618988036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,16,1,power_law_1.2,2.124550437927246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,16,balanced,0.04188266893227895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,16,1,power_law_1.2,0.1519871950149536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,16,balanced,0.03940266619126002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,16,balanced,0.039664000272750854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,16,1,power_law_1.01,0.32203519344329834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,16,balanced,0.039877332746982574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,16,1,power_law_1.2,0.16403199434280397
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,16,balanced,0.04359466830889384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,16,balanced,0.043653334180514015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,16,balanced,0.049866666396458946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,16,1,power_law_1.01,0.32499840259552004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,16,1,power_law_1.2,0.19095679521560668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,16,balanced,0.05555733541647593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,16,balanced,0.07170666754245758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,16,balanced,0.08891200025876363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,16,1,power_law_1.01,0.3345664024353027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,16,1,power_law_1.2,0.18659839630126954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,16,balanced,0.12129599849383037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,16,power_law_1.01,0.030899199843406677
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,16,balanced,0.15236266454060873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,16,1,power_law_1.2,2.973535919189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,16,1,power_law_1.01,0.3433727979660034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,16,power_law_1.01,0.031020799279212953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,16,balanced,0.21247466405232748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,16,balanced,0.2794400056203206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,16,power_law_1.01,0.03141759932041168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,16,1,power_law_1.01,0.3510656118392944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,16,power_law_1.01,0.031436800956726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,16,power_law_1.01,0.03501439988613129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,16,1,power_law_1.01,0.37360639572143556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,16,power_law_1.01,0.13125760555267335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,16,power_law_1.01,0.033932799100875856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,16,power_law_1.01,0.0351936012506485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,16,power_law_1.01,0.16280319690704345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,16,power_law_1.01,0.03765760064125061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,16,1,power_law_1.01,0.4079872131347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,16,power_law_1.01,0.03687680065631867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,16,power_law_1.01,0.2156480073928833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,16,power_law_1.01,0.037567999958992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,16,power_law_1.01,0.036422398686408994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,16,power_law_1.01,0.04032639861106872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,16,power_law_1.01,0.2730047941207886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,16,power_law_1.01,0.036345601081848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,16,power_law_1.01,0.040531200170516965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,16,power_law_1.01,0.3846719980239868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,16,power_law_1.01,0.040531200170516965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,16,power_law_1.01,0.04577920138835907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,16,power_law_1.01,0.04177919924259186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,16,1,power_law_1.2,0.2193727970123291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,16,power_law_1.01,0.04947839975357056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,16,power_law_1.01,0.5014656066894532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,16,power_law_1.01,0.04177919924259186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,16,power_law_1.01,0.05093119740486145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,16,power_law_1.01,0.05500800013542175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,16,power_law_1.01,0.054067200422286986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,16,power_law_1.01,0.7364352226257325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,16,power_law_1.01,0.05623040199279785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,16,power_law_1.01,0.05975679755210876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,16,power_law_1.01,0.0599232017993927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,16,power_law_1.01,0.06366080045700073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,16,power_law_1.01,0.05982080101966858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,16,power_law_1.01,0.9579071998596191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,16,power_law_1.01,0.0742143988609314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,16,power_law_1.01,0.06188160181045532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,16,1,power_law_1.01,0.47248001098632814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,16,1,power_law_1.2,3.871148681640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,16,power_law_1.01,0.06513280272483826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,16,1,power_law_1.01,0.5684160232543946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,16,power_law_1.01,0.07133439779281617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,16,power_law_1.2,0.031385600566864014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,16,power_law_1.01,0.07901440262794494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,16,1,power_law_1.01,0.6489535808563233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,16,power_law_1.01,0.08952959775924682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,16,1,power_law_1.2,0.23505918979644774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,16,power_law_1.01,0.11424640417099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,16,1,power_law_1.2,0.28015999794006347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,16,1,power_law_1.01,0.8485376358032226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,16,1,power_law_1.2,0.34536960124969485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,16,power_law_1.01,0.025727999210357667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,16,1,power_law_1.01,1.0472064018249512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,16,power_law_1.01,0.08837760090827942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,16,power_law_1.01,0.024985599517822265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,16,power_law_1.01,0.103603196144104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,16,power_law_1.01,0.025241601467132568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,16,1,power_law_1.01,1.4395903587341308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,16,power_law_1.01,0.026054400205612182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,16,power_law_1.2,0.03275519907474518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,16,power_law_1.2,0.03315199911594391
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,16,power_law_1.01,0.026009601354599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,16,power_law_1.2,0.034118399024009705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,16,power_law_1.01,0.02876800000667572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,16,power_law_1.2,0.032204800844192506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,16,power_law_1.01,0.031007999181747438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,16,1,power_law_1.01,1.8294336318969726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,16,power_law_1.2,0.03617919981479645
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,16,power_law_1.01,0.032128000259399415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,16,power_law_1.2,0.040300801396369934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,16,power_law_1.01,0.03298560082912445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,16,power_law_1.2,0.041631999611854556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,16,1,power_law_1.2,0.42525439262390136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,16,power_law_1.2,0.043968001008033754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,16,power_law_1.2,0.04645760059356689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,16,1,power_law_1.01,2.5759552001953123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,16,power_law_1.01,0.1344256043434143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,16,power_law_1.2,0.04992640018463135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,16,power_law_1.2,0.05369600057601929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,16,power_law_1.2,0.05434240102767944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,16,power_law_1.2,0.057548797130584715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,16,power_law_1.01,0.1377408027648926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,16,power_law_1.2,0.06371200084686279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,16,power_law_1.01,0.1806656002998352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,16,power_law_1.2,0.06776959896087646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,16,1,power_law_1.01,3.314137649536133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,16,power_law_1.01,0.23022079467773438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,16,power_law_1.2,0.07722880244255066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,16,power_law_1.01,0.03281919956207276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,16,power_law_1.01,0.32408320903778076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,16,power_law_1.01,0.03320960104465485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,16,power_law_1.01,0.41043839454650877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,16,power_law_1.01,0.049414399266242984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,16,power_law_1.01,0.047712001204490664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,16,power_law_1.01,0.1662335991859436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,16,power_law_1.01,0.6004479885101318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,16,power_law_1.01,0.03901439905166626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,16,power_law_1.01,0.2269376039505005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,16,power_law_1.01,0.038387200236320494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,16,power_law_1.01,0.772819185256958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,16,power_law_1.01,0.2895103931427002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,16,power_law_1.01,0.0383296012878418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,16,power_law_1.01,0.04015359878540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,16,power_law_1.01,0.40515837669372556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,16,power_law_1.01,1.1661952018737793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,16,power_law_1.01,0.04481280148029328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,16,power_law_1.2,0.09040639996528625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,16,power_law_1.01,0.04865280091762543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,16,power_law_1.01,0.5170495986938477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,16,power_law_1.2,0.10652159452438355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,16,1,power_law_1.2,0.4995711803436279
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,16,power_law_1.01,0.056729602813720706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,16,power_law_1.01,1.5091008186340331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,16,power_law_1.01,0.752185583114624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,16,power_law_1.01,0.06446719765663148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,16,1,power_law_1.2,0.627238416671753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,16,power_law_1.01,0.08343039751052857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,16,power_law_1.2,0.034995201230049136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,16,power_law_1.01,0.10387200117111206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,16,power_law_1.01,0.9888704299926758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,16,1,power_law_1.2,0.8675840377807618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,16,power_law_1.01,0.14103039503097534
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,16,power_law_1.01,0.17872639894485473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,16,power_law_1.2,0.029356798529624938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,16,power_law_1.01,0.25867519378662107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,16,1,power_law_1.2,1.154355239868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,16,power_law_1.2,0.030796799063682555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,16,power_law_1.01,0.32536959648132324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,16,power_law_1.2,0.031788799166679385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,16,balanced,0.03348266581694285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,16,power_law_1.2,0.13717119693756102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,16,power_law_1.2,0.02988159954547882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,16,power_law_1.2,0.16588159799575805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,16,power_law_1.2,0.03054719865322113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,16,power_law_1.2,0.035174399614334106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,16,power_law_1.2,0.033190399408340454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,16,power_law_1.2,0.025305598974227905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,16,power_law_1.2,0.03684479892253876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,16,power_law_1.2,0.02436479926109314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,16,power_law_1.2,0.037036800384521486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,16,power_law_1.2,0.025382399559020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,16,power_law_1.2,0.03686400055885315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,16,power_law_1.2,0.02635520100593567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,16,power_law_1.2,0.04123519957065582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,16,balanced,0.03507733345031738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,16,power_law_1.2,0.04193280041217804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,16,power_law_1.2,0.026310399174690247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,16,balanced,0.03375466664632162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,16,balanced,0.03944533318281174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,16,balanced,0.05596800148487091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,16,power_law_1.2,0.042412799596786496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,16,balanced,0.07961066563924153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,16,power_law_1.2,0.028697600960731505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,16,balanced,0.08380267024040222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,16,balanced,0.08912000060081482
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,16,power_law_1.2,0.031411200761795044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,16,power_law_1.2,0.032390400767326355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,16,power_law_1.2,0.03349120020866394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,16,power_law_1.2,0.2260416030883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,16,power_law_1.2,0.0382207989692688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,16,power_law_1.2,0.03319680094718933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,16,power_law_1.2,0.037945601344108584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,16,power_law_1.2,0.03311359882354736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,16,power_law_1.2,0.03986560106277466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,16,power_law_1.2,0.0489984005689621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,16,power_law_1.2,0.04028159976005554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,16,power_law_1.2,0.045875200629234315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,16,power_law_1.2,0.04901759922504425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,16,balanced,0.033813332517941795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,16,balanced,0.03605333218971888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,16,power_law_1.2,0.05572479963302612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,16,balanced,0.04168533285458883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,16,balanced,0.08981333176294963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,16,power_law_1.2,0.2848639965057373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,16,balanced,0.061941335598627724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,16,balanced,0.09706133604049683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,16,balanced,0.06427200138568878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,16,power_law_1.2,0.05492479801177978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,16,balanced,0.07939200103282928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,16,balanced,0.09805333614349365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,16,balanced,0.08040000001589458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,16,balanced,0.10325866937637329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,16,balanced,0.08043733239173889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,16,balanced,0.10778133074442546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,16,balanced,0.08177599807580312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,16,balanced,0.11117866635322571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,16,balanced,0.08552533388137817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,16,balanced,0.11560533444086711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,16,balanced,0.09146133065223694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,16,balanced,0.1211893359820048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,16,balanced,0.09476799766222636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,16,balanced,0.12756799658139548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,16,balanced,0.10180800159772237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,16,balanced,0.10707733035087585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,16,power_law_1.2,0.4085184097290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,16,balanced,0.11288000146547954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,16,balanced,0.12658666570981345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,16,balanced,0.1569973329703013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,16,balanced,0.18107734123865762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,16,balanced,0.22516266504923502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,16,balanced,0.27109332879384357
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,16,power_law_1.2,0.04633600115776062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,16,balanced,0.38236268361409503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,16,power_law_1.2,0.03860479891300202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,16,balanced,0.4708373149236043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,16,power_law_1.2,0.05098239779472351
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,16,power_law_1.2,0.04015359878540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,16,balanced,0.6690186659495035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,16,power_law_1.2,0.05454720258712768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,16,power_law_1.2,0.03918080031871796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,16,power_law_1.2,0.05699840188026428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,16,power_law_1.2,0.05799679756164551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,16,balanced,0.8655573527018229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,16,power_law_1.2,0.040531200170516965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,16,power_law_1.2,0.05866879820823669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,16,power_law_1.2,0.06530560255050659
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,16,power_law_1.2,0.04568319916725159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,16,power_law_1.2,0.06273279786109924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,16,balanced,0.13928533593813577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,16,power_law_1.2,0.07342079877853394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,16,balanced,1.259663979212443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,16,balanced,0.17037334044774374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,16,power_law_1.2,0.050860798358917235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,16,power_law_1.2,0.06678400039672852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,16,power_law_1.2,0.08910080194473266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,16,balanced,0.1904426614443461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,16,balanced,0.22977600495020548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,16,power_law_1.2,0.10658559799194336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,16,balanced,0.26999467611312866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,16,balanced,1.6534454027811687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,16,power_law_1.2,0.1370751976966858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,16,balanced,0.38629333178202313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,16,balanced,0.04601066807905833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,16,balanced,0.04756266872088114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,16,balanced,0.04808533191680908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,16,balanced,0.46601064999898273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,16,balanced,0.05764799813429514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,16,balanced,0.07228266696135204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,16,balanced,0.07976533472537994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,16,balanced,0.6617279847462972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,16,balanced,0.08307733138402303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,16,balanced,0.08212799827257793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,16,balanced,0.11206400394439697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,16,balanced,0.8500373363494873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,16,balanced,0.11275200049082439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,16,balanced,0.11767466862996419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,16,balanced,0.12005866567293803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,16,balanced,0.12550399700800577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,16,balanced,1.2402666409810383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,16,power_law_1.2,0.0738431990146637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,16,power_law_1.2,0.0807424008846283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,16,power_law_1.2,0.09409919977188111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,16,power_law_1.2,0.16592639684677124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,16,power_law_1.2,0.11544320583343506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,16,power_law_1.2,0.2334144115447998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,16,power_law_1.2,0.1474560022354126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,16,power_law_1.2,0.18691200017929077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,16,power_law_1.2,0.24506239891052245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,16,balanced,0.13425599535306296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,16,balanced,0.141077329715093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,16,power_law_1.2,0.346560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,16,1,balanced,0.04346133271853129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,16,balanced,1.6289119720458984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,16,power_law_1.2,0.4479167938232422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,16,power_law_1.2,0.2949183940887451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,16,balanced,0.1476533313592275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,16,balanced,0.16639467080434164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,16,balanced,0.199237326780955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,16,power_law_1.2,0.5330048084259034
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,16,power_law_1.2,0.05719040036201477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,16,balanced,0.23533866802851358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,16,balanced,0.33480532964070636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,16,balanced,0.4008266528447469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,16,balanced,0.5703039964040121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,16,power_law_1.2,0.6334976196289063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,16,balanced,0.7334720293680826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,16,balanced,1.0660160382588704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,16,balanced,0.02773866554101308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,16,balanced,0.027632000545660656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,16,balanced,1.395487944285075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,16,balanced,0.027424000203609467
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,16,balanced,0.029557332396507263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,16,balanced,0.03284800052642822
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,16,balanced,0.033904001116752625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,16,balanced,0.03872533390919367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,16,balanced,0.04067199925581614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,16,power_law_1.2,0.06844800114631652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,16,balanced,0.059104000528653465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,16,balanced,2.046720027923584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,16,balanced,0.06322666505972545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,16,power_law_1.2,0.08739839792251587
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,16,balanced,0.05684266487757365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,16,balanced,0.12043199936548869
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,16,balanced,0.11449600259462993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,16,power_law_1.2,0.11144959926605225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,16,balanced,0.09435733159383138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,16,balanced,0.06379200021425883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,16,balanced,0.0661599983771642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,16,balanced,2.7025492986043296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,16,power_law_1.2,0.14871679544448851
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,16,balanced,0.06005866825580597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,16,balanced,0.06803733110427856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,16,balanced,0.07262933254241943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,16,power_law_1.01,0.07786880135536194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,16,balanced,0.08678399523099263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,16,balanced,0.09904000163078308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,16,power_law_1.01,0.07793279886245727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,16,balanced,0.12997866670290628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,16,balanced,0.15414399902025858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,16,power_law_1.01,0.07827200293540955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,16,balanced,0.2133973240852356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,16,power_law_1.01,0.0728767991065979
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,16,balanced,0.27713600794474286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,16,balanced,0.39257065455118817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,16,balanced,0.5088320175806681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,16,power_law_1.01,0.05437440276145935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,16,power_law_1.01,0.05777919888496399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,16,power_law_1.01,0.058412802219390866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,16,power_law_1.01,0.05264000296592712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,16,power_law_1.01,0.05432320237159729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,16,power_law_1.01,0.0597760021686554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,16,power_law_1.01,0.07587199807167053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,16,power_law_1.01,0.07512320280075073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,16,power_law_1.01,0.08025599718093872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,16,power_law_1.01,0.07689599990844727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,16,power_law_1.01,0.08372480273246766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,16,power_law_1.01,0.07650560140609741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,16,power_law_1.01,0.08451840281486511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,16,power_law_1.01,0.07850880026817322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,16,power_law_1.01,0.09581440091133117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,16,power_law_1.01,0.08371840119361877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,16,power_law_1.01,0.09550719857215881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,16,power_law_1.01,0.08794879913330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,16,power_law_1.01,0.10168960094451904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,16,power_law_1.01,0.09322239756584168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,16,power_law_1.01,0.10539519786834717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,16,power_law_1.01,0.09971200227737427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,16,power_law_1.01,0.10867199897766114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,16,power_law_1.01,0.1072767972946167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,16,1,balanced,0.04990399877230326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,16,power_law_1.2,0.8617280006408692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,16,power_law_1.01,0.11851520538330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,16,power_law_1.01,0.13435519933700563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,16,power_law_1.2,0.41775360107421877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,16,power_law_1.01,0.1653439998626709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,16,power_law_1.01,0.19726719856262206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,16,power_law_1.01,0.24849920272827147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,16,power_law_1.01,0.3020927906036377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,16,power_law_1.01,0.40988798141479493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,16,power_law_1.01,0.11661440134048462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,16,1,balanced,0.05906666815280914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,16,1,balanced,0.07644799848397572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,16,power_law_1.01,0.5164671897888183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,16,power_law_1.01,0.7299839973449707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,16,power_law_1.01,0.9580415725708008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,16,power_law_1.01,1.4045632362365723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,16,power_law_1.01,0.1251263976097107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,16,power_law_1.01,0.13445760011672975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,16,power_law_1.01,0.15084160566329957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,16,power_law_1.01,1.8704767227172852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,16,power_law_1.01,0.19139839410781861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,16,power_law_1.01,0.06441599726676941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,16,power_law_1.01,0.22453761100769043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,16,power_law_1.01,0.06544640064239501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,16,power_law_1.01,0.28008959293365476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,16,power_law_1.01,0.06496000289916992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,16,power_law_1.01,0.34024319648742674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,16,power_law_1.01,0.06680319905281067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,16,power_law_1.01,0.4595200061798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,16,power_law_1.01,0.0650111973285675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,16,1,power_law_1.2,0.04154239892959595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,16,power_law_1.01,0.07593600153923034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,16,power_law_1.01,0.5839424133300781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,16,power_law_1.01,0.07810559868812561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,16,power_law_1.01,0.07925119996070862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,16,power_law_1.01,0.8031167984008789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,16,power_law_1.01,0.10755840539932252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,16,power_law_1.01,0.10921599864959716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,16,power_law_1.01,0.11375999450683594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,16,power_law_1.01,1.1042367935180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,16,power_law_1.01,0.11741440296173096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,16,power_law_1.01,0.12286720275878907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,16,power_law_1.01,0.13559679985046386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,16,power_law_1.01,1.526591968536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,16,power_law_1.01,0.15073280334472655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,16,power_law_1.2,0.5487040042877197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,16,power_law_1.01,0.16496000289916993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,16,power_law_1.01,0.1936959981918335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,16,power_law_1.01,2.0239679336547853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,16,power_law_1.01,0.23912320137023926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,16,power_law_1.01,0.2879103899002075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,16,power_law_1.01,0.39761919975280763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,16,power_law_1.01,0.4860352039337158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,16,1,balanced,0.11289599537849426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,16,power_law_1.01,0.6866047859191895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,16,power_law_1.01,0.9492671966552735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,16,1,power_law_1.2,0.046003198623657225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,16,1,power_law_1.2,0.06104959845542908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,16,power_law_1.01,1.3931520462036133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,16,1,power_law_1.2,0.0742464005947113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,16,power_law_1.2,1.3018367767333985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,16,power_law_1.01,1.713248062133789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,16,power_law_1.01,0.03268480002880096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,16,power_law_1.01,0.032576000690460204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,16,power_law_1.01,0.03264000117778778
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,16,power_law_1.01,0.03224320113658905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,16,power_law_1.01,2.4020736694335936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,16,power_law_1.2,0.19866880178451538
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,16,power_law_1.01,0.03207040131092072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,16,power_law_1.01,0.034918400645256045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,16,power_law_1.01,0.03749760091304779
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,16,power_law_1.01,0.04051840007305145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,16,power_law_1.01,0.05694720149040222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,16,power_law_1.01,3.2838016510009767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,16,power_law_1.01,0.057030397653579715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,16,power_law_1.2,0.07448959946632386
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,16,power_law_1.01,0.05059199929237366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,16,power_law_1.2,0.7505727767944336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,16,power_law_1.01,0.11271040439605713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,16,1,balanced,0.17987199624379477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,16,1,balanced,0.18207999070485434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,16,power_law_1.01,0.10057599544525146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,16,1,balanced,0.1821813384691874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,16,power_law_1.01,0.07868160009384155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,16,power_law_1.01,0.061312001943588254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,16,power_law_1.01,0.06233599781990051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,16,power_law_1.01,0.06394879817962647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,16,power_law_1.01,0.06933119893074036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,16,power_law_1.01,0.07869439721107482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,16,power_law_1.2,0.07095040082931518
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,16,power_law_1.01,0.09562240242958069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,16,power_law_1.2,0.07529600262641907
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,16,power_law_1.01,0.11367039680480957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,16,power_law_1.2,0.0654911994934082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,16,power_law_1.01,0.1463871955871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,16,power_law_1.2,0.06657919883728028
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,16,power_law_1.01,0.18877439498901366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,16,power_law_1.2,0.07496320009231568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,16,power_law_1.2,0.08218880295753479
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,16,power_law_1.01,0.26213760375976564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,16,power_law_1.2,0.08373759984970093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,16,power_law_1.01,0.31731200218200684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,16,power_law_1.2,0.09423999786376953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,16,power_law_1.2,0.09689599871635438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,16,power_law_1.01,0.4659264087677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,16,power_law_1.2,0.09870719909667969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,16,power_law_1.2,0.10478719472885131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,16,power_law_1.01,0.6447872161865235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,16,power_law_1.2,0.11093120574951172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,16,power_law_1.2,0.0543615996837616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,16,power_law_1.2,0.11769599914550781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,16,power_law_1.2,0.054681599140167236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,16,power_law_1.2,0.12563199996948243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,16,power_law_1.2,0.05978239774703979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,16,1,power_law_1.2,0.09258880019187928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,16,power_law_1.2,0.13697919845581055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,16,power_law_1.2,0.05687040090560913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,16,power_law_1.2,0.15734399557113649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,16,power_law_1.2,0.05658239722251892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,16,power_law_1.2,0.20583679676055908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,16,power_law_1.2,0.061478400230407716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,16,power_law_1.2,0.23410561084747314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,16,power_law_1.2,0.07444480061531067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,16,power_law_1.2,0.2869632005691528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,16,power_law_1.2,0.0766592025756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,16,power_law_1.2,0.07803519964218139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,16,power_law_1.2,0.07875840067863464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,16,power_law_1.2,0.3542720079421997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,16,power_law_1.2,0.08416000008583069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,16,power_law_1.2,0.08883839845657349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,16,power_law_1.2,0.09340159893035889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,16,power_law_1.2,0.09996799826622009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,16,power_law_1.2,0.11244800090789794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,16,power_law_1.2,0.11916160583496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,16,power_law_1.2,0.7904831886291503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,16,power_law_1.2,0.14001920223236083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,16,power_law_1.2,0.17070080041885377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,16,power_law_1.2,0.19875839948654175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,16,power_law_1.2,0.25597438812255857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,16,power_law_1.2,0.5127232074737549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,16,1,balanced,0.3371573289235433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,16,1,balanced,0.3387146790822347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,16,power_law_1.2,0.6322175979614257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,16,power_law_1.2,0.3075839996337891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,16,1,balanced,0.3447786569595337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,16,1,balanced,0.34771732489267987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,16,power_law_1.2,0.4266816139221191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,16,1,balanced,0.35099732875823975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,16,1,balanced,0.3564586639404297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,16,1,balanced,0.3629386822382609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,16,1,balanced,0.37143464883168537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,16,1,power_law_1.2,0.11393920183181763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,16,1,balanced,0.3820426861445109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,16,1,balanced,0.40447998046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,16,1,power_law_1.2,0.14113279581069946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,16,1,balanced,0.4280266761779785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,16,1,balanced,0.47447999318440753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,16,1,power_law_1.2,0.14695039987564087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,16,1,balanced,0.5235306819279989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,16,1,balanced,0.8332640329996744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,16,1,balanced,0.9266613324483236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,16,power_law_1.2,0.5544832229614258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,16,1,balanced,1.334778626759847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,16,1,balanced,1.6994560559590657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,16,power_law_1.2,0.8913791656494141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,16,1,balanced,2.5023306210835776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,16,power_law_1.2,1.1575872421264648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,16,1,balanced,3.271594683329264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,16,power_law_1.2,0.06477439999580384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,16,power_law_1.2,0.8163392066955566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,16,power_law_1.2,0.06669440269470214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,16,power_law_1.2,0.2638911962509155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,16,power_law_1.2,0.06602879762649536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,16,power_law_1.2,0.06327040195465088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,16,power_law_1.2,0.06025599837303162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,16,power_law_1.2,0.07631999850273133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,16,power_law_1.2,0.07741439938545228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,16,power_law_1.2,0.08047999739646912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,16,power_law_1.2,0.10933120250701904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,16,power_law_1.2,0.10871679782867431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,16,power_law_1.2,0.1136896014213562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,16,power_law_1.2,0.1187391996383667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,16,power_law_1.2,0.12575360536575317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,16,power_law_1.2,0.1382464051246643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,16,power_law_1.2,0.15277440547943116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,16,power_law_1.2,0.16799999475479127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,16,power_law_1.2,0.2022655963897705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,16,power_law_1.2,0.2551104068756104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,16,power_law_1.2,0.31640961170196535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,16,power_law_1.2,0.40808959007263185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,16,power_law_1.2,0.5528448104858399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,16,power_law_1.2,0.7503551959991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,16,1,power_law_1.2,0.3019007921218872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,16,1,power_law_1.2,0.3059776067733765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,16,power_law_1.2,1.0167296409606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,16,power_law_1.2,1.0453503608703614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,16,1,power_law_1.2,0.3141439914703369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,16,1,power_law_1.2,0.32113919258117674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,16,1,power_law_1.2,0.3272511959075928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,16,1,power_law_1.2,0.33834240436553953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,16,1,power_law_1.2,0.35066239833831786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,16,1,power_law_1.2,0.35696001052856446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,16,power_law_1.2,1.4477952003479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,16,1,power_law_1.2,0.3798271894454956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,16,1,power_law_1.2,0.41987199783325196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,16,power_law_1.2,1.5903039932250977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,16,1,power_law_1.2,0.5029056072235107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,16,1,power_law_1.2,0.5814464092254639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,16,1,power_law_1.2,0.7011775970458984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,16,1,power_law_1.2,0.863929557800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,16,power_law_1.2,1.8277631759643556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,16,1,power_law_1.2,1.0547648429870606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,16,1,power_law_1.2,1.4506367683410644
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,16,power_law_1.2,0.0304639995098114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,16,1,power_law_1.2,1.8375680923461915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,16,power_law_1.2,0.029548799991607665
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,16,power_law_1.2,0.029919999837875366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,16,power_law_1.2,0.030355200171470642
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,16,power_law_1.2,0.030297601222991945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,16,1,power_law_1.2,2.57324161529541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,16,power_law_1.2,0.03238399922847748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,16,power_law_1.2,0.03731200098991394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,16,power_law_1.2,0.03687039911746979
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,16,power_law_1.2,0.05315200090408325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,16,power_law_1.2,0.05175039768218994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,16,1,power_law_1.2,3.3315967559814452
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,16,power_law_1.2,0.04425599873065948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,16,1,balanced,0.02922666569550832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,16,1,balanced,0.033573334415753685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,16,1,balanced,0.041759997606277466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,16,1,balanced,0.07363733152548473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,16,power_law_1.2,0.09041919708251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,16,1,balanced,0.11796266833941142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,16,1,balanced,0.11798933148384094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,16,1,balanced,0.11245333154996236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,16,power_law_1.2,0.07789440155029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,16,1,balanced,0.11308800180753072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,16,1,balanced,0.10319999853769939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,16,1,balanced,0.10462400317192078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,16,power_law_1.2,0.057043200731277464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,16,1,balanced,0.11876266201337178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,16,1,balanced,0.11752532919247945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,16,1,balanced,0.11967999736467998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,16,power_law_1.2,0.07221760153770447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,16,1,balanced,0.12215999762217204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,16,1,balanced,0.12590400377909342
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,16,power_law_1.2,0.066048002243042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,16,1,balanced,0.1308693289756775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,16,1,balanced,0.13910399874051413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,16,1,balanced,0.1736746629079183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,16,power_law_1.2,0.06324480175971985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,16,1,balanced,0.1868106722831726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,16,1,balanced,0.2118133306503296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,16,power_law_1.2,0.07349759936332703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,16,1,balanced,0.2366559902826945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,16,1,balanced,0.35436801115671795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,16,power_law_1.2,0.08158079981803894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,16,1,balanced,0.4052586555480957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,16,power_law_1.2,0.09679359793663025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,16,1,balanced,0.5761546691258749
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,16,power_law_1.2,0.11288959980010986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,16,1,balanced,0.7515040238698324
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,16,power_law_1.2,0.1496384024620056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,16,1,balanced,1.1061333020528157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,16,1,balanced,1.4468107223510742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,16,1,balanced,0.02628266563018163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,16,1,balanced,0.027461332579453785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,16,1,balanced,0.03138133386770884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,16,1,balanced,0.0460746685663859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,16,1,balanced,0.07228266696135204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,16,1,balanced,0.07434666653474171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,16,1,balanced,0.08854400118192036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,16,1,balanced,0.08669333656628926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,16,1,balanced,0.08754666646321614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,16,1,balanced,0.08893866340319316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,16,1,balanced,0.09913600484530131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,16,1,balanced,0.10111467043558757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,16,1,balanced,0.10259200135866801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,16,1,balanced,0.10528533657391866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,16,1,balanced,0.11117866635322571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,16,1,balanced,0.11421333750089009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,16,1,balanced,0.12196266651153564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,16,power_law_1.2,1.7344703674316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,16,power_law_1.2,0.19775359630584716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,16,1,balanced,0.15582933028539023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,16,1,balanced,0.16921067237854004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,16,1,balanced,0.1978613336881002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,16,1,balanced,0.22626133759816489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,16,1,balanced,0.333679993947347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,16,1,balanced,0.3874719937642415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,16,1,balanced,0.552346666653951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,16,1,balanced,0.7167413234710693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,16,1,balanced,1.0513013203938801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,16,1,balanced,1.3784480094909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,16,1,power_law_1.01,0.0281792014837265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,16,1,power_law_1.01,0.03366400003433227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,16,1,power_law_1.01,0.04271360039710999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,16,1,power_law_1.01,0.060102397203445436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,16,1,power_law_1.01,0.07348480224609374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,16,1,power_law_1.01,0.10161279439926148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,16,1,power_law_1.01,0.10432640314102173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,16,1,power_law_1.01,0.10611200332641602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,16,1,power_law_1.01,0.09830399751663207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,16,1,power_law_1.01,0.09946879744529724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,16,1,power_law_1.01,0.11263999938964844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,16,power_law_1.2,1.4922752380371094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,16,1,power_law_1.01,0.11340800523757935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,16,1,power_law_1.01,0.11709439754486084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,16,1,power_law_1.01,0.121343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,16,1,power_law_1.01,0.12538880109786987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,16,1,power_law_1.01,0.1314687967300415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,16,power_law_1.2,0.2983551979064941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,16,1,power_law_1.01,0.14328960180282593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,16,1,power_law_1.01,0.17688319683074952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,16,1,power_law_1.01,0.22515199184417725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,16,1,power_law_1.01,0.2609087944030762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,16,1,power_law_1.01,0.29895040988922117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,16,1,power_law_1.01,0.36942079067230227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,16,1,power_law_1.01,0.4688511848449707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,16,power_law_1.2,0.37059199810028076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,16,1,power_law_1.01,0.6469183921813965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,16,1,power_law_1.01,0.825376033782959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,16,power_law_1.2,0.5156223773956299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,16,1,power_law_1.01,1.1617471694946289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,16,1,power_law_1.01,1.525984001159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,16,1,power_law_1.01,0.023443199694156647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,16,power_law_1.2,0.3472192049026489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,16,1,power_law_1.01,0.026495999097824095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,16,1,power_law_1.01,0.034220799803733826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,16,1,power_law_1.2,0.028204798698425293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,16,power_law_1.2,2.7600704193115235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,16,1,power_law_1.01,0.042656001448631284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,16,1,power_law_1.01,0.04973439872264862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,16,1,power_law_1.01,0.06670079827308655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,16,1,power_law_1.01,0.08110719919204712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,16,1,power_law_1.01,0.08229119777679443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,16,1,power_law_1.01,0.0839680016040802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,16,1,power_law_1.01,0.0856000006198883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,16,1,power_law_1.01,0.0946175992488861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,16,1,power_law_1.01,0.09711999893188476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,16,1,power_law_1.01,0.10000640153884888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,16,1,power_law_1.2,0.03498879969120026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,16,1,power_law_1.01,0.10409599542617798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,16,1,power_law_1.2,0.04286719858646393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,16,1,power_law_1.01,0.10982400178909302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,16,1,power_law_1.2,0.05598719716072083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,16,1,power_law_1.01,0.11512320041656494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,16,1,power_law_1.2,0.06995199918746949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,16,1,power_law_1.01,0.1256832003593445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,16,1,power_law_1.2,0.10374399423599243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,16,1,power_law_1.01,0.16343679428100585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,16,1,power_law_1.2,0.1049280047416687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,16,1,power_law_1.01,0.19549440145492553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,16,1,power_law_1.2,0.10618239641189575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,16,1,power_law_1.01,0.2338560104370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,16,1,power_law_1.2,0.09867519736289979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,16,1,power_law_1.01,0.27022719383239746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,16,1,power_law_1.2,0.10090880393981934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,16,1,power_law_1.2,0.11415679454803467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,16,1,power_law_1.01,0.34215679168701174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,16,1,power_law_1.2,0.11414400339126587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,16,1,power_law_1.01,0.4341887950897217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,16,1,power_law_1.2,0.11758079528808593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,16,1,power_law_1.2,0.12259839773178101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,16,1,power_law_1.01,0.600767993927002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,16,1,power_law_1.2,0.12706559896469116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,16,1,power_law_1.2,0.13363840579986572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,16,1,power_law_1.01,0.7667456150054932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,16,1,power_law_1.2,0.145632004737854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,16,1,power_law_1.2,0.20325760841369628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,16,1,power_law_1.01,1.0950207710266113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,16,1,power_law_1.2,0.22770559787750244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,16,1,power_law_1.2,0.2662911891937256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,16,1,power_law_1.01,1.4327360153198243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,16,1,power_law_1.2,0.3021183967590332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,16,1,power_law_1.2,0.023827199637889863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,16,1,power_law_1.2,0.026976001262664796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,16,1,power_law_1.2,0.03451519906520843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,16,1,power_law_1.2,0.04184960126876831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,16,1,power_law_1.2,0.04993920028209686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,16,1,power_law_1.2,0.3750335931777954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,16,1,power_law_1.2,0.06713600158691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,16,1,power_law_1.2,0.08190720081329346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,16,1,power_law_1.2,0.08245760202407837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,16,1,power_law_1.2,0.08439040184020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,16,1,power_law_1.2,0.08636159896850586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,16,1,power_law_1.2,0.09550719857215881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,16,1,power_law_1.2,0.09754239916801452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,16,1,power_law_1.2,0.10152959823608398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,16,1,power_law_1.2,0.10609920024871826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,16,1,power_law_1.2,0.11088639497756958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,16,1,power_law_1.2,0.11624319553375244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,16,1,power_law_1.2,0.12744319438934326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,16,1,power_law_1.2,0.17493760585784912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,16,1,power_law_1.2,0.19860479831695557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,16,1,power_law_1.2,0.47278079986572263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,16,1,power_law_1.2,0.2358783960342407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,16,1,power_law_1.2,0.27203199863433836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,16,1,power_law_1.2,0.348140811920166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,16,1,power_law_1.2,0.6507199764251709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,16,power_law_1.2,0.9748736381530761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,16,balanced,0.04788800080617269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,16,balanced,0.0476746658484141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,16,balanced,0.0639626681804657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,16,balanced,0.08634666601816814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,16,balanced,0.15267200271288553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,16,power_law_1.2,1.022214412689209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,16,balanced,0.18171733617782593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,16,balanced,0.03793066740036011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,16,balanced,0.039674667020638786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,16,balanced,0.04791999856630961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,16,balanced,0.070592001080513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,16,balanced,0.10389332969983418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,16,balanced,0.12546666463216147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,16,balanced,0.137472003698349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,16,balanced,0.14154133200645447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,16,balanced,0.14851199587186178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,16,balanced,0.14727466305096945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,16,balanced,0.1824693282445272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,16,balanced,0.19328532616297403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,16,balanced,0.19882132609685263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,16,balanced,0.2097439964612325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,16,balanced,0.21824000279108682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,16,balanced,0.2307466665903727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,16,balanced,0.2527466615041097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,16,balanced,0.28551467259724933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,16,balanced,0.3216266632080078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,16,balanced,0.20807466904322305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,16,balanced,0.3873279889424642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,16,balanced,0.20928533871968588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,16,balanced,0.1968266765276591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,16,balanced,0.45761601130167645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,16,balanced,0.19817066192626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,16,1,power_law_1.2,0.438643217086792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,16,balanced,0.2363146742184957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,16,balanced,0.6478399833043417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,16,balanced,0.24764800071716309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,16,balanced,0.25199999411900836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,16,balanced,0.7833546797434489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,16,balanced,0.2623680035273234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,16,balanced,0.27198400100072223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,16,balanced,0.28309865792592365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,16,balanced,1.111466646194458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,16,balanced,0.3044319947560628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,16,balanced,0.34592000643412274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,16,balanced,0.37666134039560956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,16,balanced,1.425989309946696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,16,balanced,0.4357120196024577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,16,balanced,0.49829332033793133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,16,balanced,0.7153813044230143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,16,balanced,2.025269349416097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,16,balanced,0.8409653504689535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,16,balanced,1.19595201810201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,16,balanced,2.668176015218099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,16,1,power_law_1.2,0.6057983875274658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,16,balanced,1.5310880343119304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,16,balanced,0.058245331048965454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,16,balanced,0.058143998185793556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,16,balanced,0.07320000231266022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,16,balanced,0.08530132969220479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,16,balanced,0.10867733756701152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,16,balanced,0.1246666709582011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,16,balanced,0.12478933731714885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,16,balanced,2.1629066467285156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,16,balanced,0.12868266304334006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,16,balanced,0.23466666539510092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,16,balanced,0.2365493377049764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,16,balanced,0.24518932898839316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,16,balanced,0.2502826650937398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,16,balanced,0.25811199347178143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,16,balanced,0.27194132407506305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,16,balanced,2.8700478871663413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,16,balanced,0.28329066435496014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,16,balanced,0.296015997727712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,16,balanced,0.32228267192840576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,16,balanced,0.3753439982732137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,16,balanced,0.4283359845479329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,16,balanced,0.6293546756108602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,16,balanced,0.734773317972819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,16,balanced,1.0464159647623699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,16,balanced,1.346826712290446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,16,balanced,1.862533410390218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,16,balanced,2.468128045399984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,16,balanced,3.590976079305013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,16,balanced,4.706101417541504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,16,balanced,0.031258667508761086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,16,balanced,0.0313226655125618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,16,balanced,0.031541332602500916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,16,balanced,0.04369066655635834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,16,power_law_1.01,0.14476799964904785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,16,balanced,0.06500266492366791
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,16,balanced,0.07353066901365916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,16,balanced,0.07635733485221863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,16,power_law_1.01,0.16725120544433594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,16,balanced,0.07799999912579854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,16,balanced,0.07675200204054515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,16,power_law_1.01,0.1711680054664612
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,16,balanced,0.11424533526102702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,16,balanced,0.10550933082898457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,16,balanced,0.09870400031407674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,16,power_law_1.01,0.16751359701156615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,16,balanced,0.2712799906730652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,16,power_law_1.01,0.13215999603271483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,16,power_law_1.01,0.16403839588165284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,16,power_law_1.01,0.19630719423294068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,16,power_law_1.01,0.20208640098571778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,16,power_law_1.01,0.19349119663238526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,16,power_law_1.01,0.18428159952163697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,16,power_law_1.01,0.22006399631500245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,16,power_law_1.01,0.22522239685058593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,16,balanced,0.2093706727027893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,16,balanced,0.17349867026011148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,16,balanced,0.13272000352541605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,16,balanced,0.10494933525721233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,16,balanced,0.1069599986076355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,16,balanced,0.11587733030319214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,16,balanced,0.1344533363978068
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,16,balanced,0.15214932958285013
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,16,balanced,0.20921599864959717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,16,balanced,0.24714666604995728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,16,balanced,0.35022934277852374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,16,balanced,0.45431466897328693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,16,power_law_1.01,0.23503999710083007
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,16,balanced,0.6254400014877319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,16,power_law_1.01,0.24614400863647462
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,16,balanced,0.8807413578033447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,16,power_law_1.01,0.281279993057251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,16,power_law_1.01,0.09294080138206481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,16,power_law_1.01,0.2919487953186035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,16,power_law_1.01,0.11799039840698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,16,power_law_1.01,0.3072448015213013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,16,power_law_1.01,0.11419520378112794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,16,1,power_law_1.2,0.7753983974456787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,16,power_law_1.01,0.11978880167007447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,16,1,power_law_1.2,0.8281984329223633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,16,power_law_1.2,2.1104703903198243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,16,power_law_1.01,0.3682240009307861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,16,power_law_1.01,0.42257280349731446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,16,power_law_1.01,0.09253759980201721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,16,power_law_1.01,0.11847679615020752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,16,power_law_1.01,0.524729585647583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,16,power_law_1.01,0.1323456048965454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,16,power_law_1.01,0.637830400466919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,16,power_law_1.01,0.1341055989265442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,16,power_law_1.01,0.1353600025177002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,16,power_law_1.01,0.864192008972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,16,power_law_1.01,0.14289920330047606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,16,power_law_1.01,0.17503999471664428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,16,power_law_1.01,1.0293824195861816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,16,power_law_1.01,0.1898687958717346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,16,power_law_1.01,0.10226559638977051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,16,power_law_1.01,0.19302400350570678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,16,power_law_1.01,0.11552000045776367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,16,power_law_1.01,0.20274560451507567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,16,power_law_1.01,1.5510784149169923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,16,power_law_1.01,0.10750080347061157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,16,power_law_1.01,0.22121601104736327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,16,power_law_1.01,0.11699199676513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,16,power_law_1.01,0.22657279968261718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,16,power_law_1.01,0.09986559748649597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,16,power_law_1.01,0.2546303987503052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,16,power_law_1.01,2.0601408004760744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,16,power_law_1.01,0.117740797996521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,16,power_law_1.01,0.3002943992614746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,16,power_law_1.2,0.7607039928436279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,16,power_law_1.01,0.12244479656219483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,16,power_law_1.01,0.3512511968612671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,16,power_law_1.01,0.1253056049346924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,16,power_law_1.01,0.2310528039932251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,16,power_law_1.01,0.4383999824523926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,16,power_law_1.01,0.23208320140838623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,16,power_law_1.01,2.8121023178100586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,16,power_law_1.01,0.5265088081359863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,16,power_law_1.01,0.24065918922424318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,16,power_law_1.01,0.24571518898010253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,16,power_law_1.01,0.7209216117858886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,16,power_law_1.01,0.25569279193878175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,16,power_law_1.01,0.2699392080307007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,16,power_law_1.01,0.8784319877624511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,16,power_law_1.01,0.2931391954421997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,16,power_law_1.01,3.6954113006591798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,16,power_law_1.01,0.3099263906478882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,16,power_law_1.2,0.1448192000389099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,16,power_law_1.01,0.37214720249176025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,16,power_law_1.2,0.1628224015235901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,16,power_law_1.2,0.16469759941101075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,16,power_law_1.2,0.16841599941253663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,16,power_law_1.2,0.12152320146560669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,16,power_law_1.2,0.17235200405120848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,16,power_law_1.2,0.19512319564819336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,16,power_law_1.2,0.1925503969192505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,16,power_law_1.01,1.291635227203369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,16,power_law_1.2,0.19288320541381837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,16,power_law_1.2,0.18606079816818238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,16,power_law_1.01,0.46175360679626465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,16,power_law_1.2,0.2249216079711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,16,1,power_law_1.2,1.168620777130127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,16,power_law_1.01,1.6111679077148438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,16,power_law_1.01,0.5437568187713623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,16,power_law_1.01,0.058329600095748904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,16,power_law_1.01,0.06420480012893677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,16,power_law_1.01,0.7239232063293457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,16,power_law_1.01,0.05863680243492127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,16,power_law_1.01,2.4162559509277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,16,power_law_1.01,0.06311039924621582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,16,power_law_1.01,0.9137215614318848
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,16,power_law_1.01,0.05663999915122986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,16,power_law_1.01,0.06863999962806702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,16,power_law_1.01,1.2912320137023925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,16,power_law_1.01,0.06926079988479614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,16,power_law_1.2,0.23405439853668214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,16,power_law_1.01,0.07183359861373902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,16,power_law_1.01,3.187084770202637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,16,power_law_1.2,0.24756479263305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,16,power_law_1.01,0.07408000230789184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,16,power_law_1.01,1.6914495468139648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,16,power_law_1.2,0.09305599927902222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,16,power_law_1.01,0.1010815978050232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,16,power_law_1.2,0.27866880893707274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,16,power_law_1.2,0.0904640018939972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,16,power_law_1.01,0.09097599983215332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,16,power_law_1.2,0.2822016000747681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,16,power_law_1.2,0.11646720170974731
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,16,power_law_1.01,0.08366720080375671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,16,power_law_1.2,0.2975935935974121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,16,power_law_1.2,0.11681920289993286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,16,power_law_1.01,0.1890944004058838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,16,power_law_1.01,2.54201602935791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,16,power_law_1.2,0.3176896095275879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,16,power_law_1.2,0.09840000271797181
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,16,power_law_1.01,0.1700096011161804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,16,power_law_1.2,0.3911168098449707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,16,power_law_1.2,0.10427520275115967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,16,power_law_1.01,0.12707200050354003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,16,power_law_1.2,0.1303104043006897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,16,power_law_1.2,0.4468095779418945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,16,power_law_1.2,0.12929919958114625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,16,power_law_1.2,0.5531648159027099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,16,power_law_1.01,3.292108917236328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,16,power_law_1.2,0.6774975776672363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,16,power_law_1.2,0.8618751525878906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,16,power_law_1.01,0.08984959721565247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,16,power_law_1.01,4.841241455078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,16,power_law_1.01,0.10094079971313477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,16,power_law_1.2,0.13242239952087403
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,16,power_law_1.01,0.11102080345153809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,16,power_law_1.2,0.14099839925765992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,16,power_law_1.01,0.12637439966201783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,16,power_law_1.2,0.17319040298461913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,16,power_law_1.2,0.17412480115890502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,16,power_law_1.01,0.15159039497375487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,16,power_law_1.2,0.19140479564666749
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,16,power_law_1.01,0.18952319622039795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,16,power_law_1.2,0.20853760242462158
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,16,power_law_1.01,0.24348800182342528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,16,power_law_1.2,1.2278528213500977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,16,power_law_1.2,0.2216576099395752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,16,power_law_1.01,0.31146879196166993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,16,power_law_1.2,0.2300800085067749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,16,power_law_1.2,1.9572223663330077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,16,power_law_1.01,6.166169738769531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,16,power_law_1.2,0.2559295892715454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,16,power_law_1.01,0.421676778793335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,16,power_law_1.2,0.10351359844207764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,16,power_law_1.01,0.5579135894775391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,16,power_law_1.2,0.10703359842300415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,16,power_law_1.2,0.1134719967842102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,16,power_law_1.2,1.7072832107543945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,16,power_law_1.01,0.8142975807189942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,16,power_law_1.2,0.11968640089035035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,16,power_law_1.2,0.10116480588912964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,16,power_law_1.2,0.11730560064315795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,16,power_law_1.01,1.1774656295776367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,16,1,balanced,0.03736533224582672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,16,power_law_1.2,0.12167680263519287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,16,1,balanced,0.049216002225875854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,16,1,balanced,0.07217066486676534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,16,1,balanced,0.11959999799728394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,16,power_law_1.2,0.12525440454483033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,16,1,balanced,0.21866132815678915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,16,power_law_1.2,0.3025791883468628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,16,power_law_1.2,0.22850561141967773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,16,power_law_1.2,0.34985599517822263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,16,power_law_1.2,0.23255679607391358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,16,power_law_1.2,0.24001920223236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,16,power_law_1.2,0.45895681381225584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,16,power_law_1.2,0.24682240486145018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,16,power_law_1.2,0.542630386352539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,16,power_law_1.2,0.25807359218597414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,16,power_law_1.2,0.27326719760894774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,16,power_law_1.2,0.7320767879486084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,16,power_law_1.2,0.3053056001663208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,16,1,balanced,0.26073066393534344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,16,power_law_1.2,0.9697983741760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,16,power_law_1.2,0.327839994430542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,16,power_law_1.2,0.38065280914306643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,16,power_law_1.2,1.3088383674621582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,16,power_law_1.2,0.46596479415893555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,16,power_law_1.2,0.588915205001831
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,16,power_law_1.2,0.05926399827003479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,16,power_law_1.2,1.850169563293457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,16,power_law_1.2,0.05917440056800842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,16,power_law_1.2,0.841209602355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,16,power_law_1.2,0.06100479960441589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,16,1,balanced,0.2783733407656352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,16,power_law_1.2,0.06354560256004334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,16,power_law_1.2,1.023033618927002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,16,power_law_1.2,0.06343039870262146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,16,power_law_1.2,3.4980094909667967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,16,power_law_1.2,0.06785280108451844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,16,1,balanced,0.030106666187445324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,16,power_law_1.2,1.4188480377197266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,16,1,balanced,0.035317334036032356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,16,1,balanced,0.05091733237107595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,16,power_law_1.2,0.06600319743156433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,16,power_law_1.2,0.07349119782447815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,16,1,power_law_1.2,1.1086400032043457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,16,power_law_1.2,1.8309312820434571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,16,1,balanced,0.2797333399454753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,16,1,balanced,0.2144533395767212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,16,1,balanced,0.21345067024230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,16,1,balanced,0.26613332827885944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,16,1,balanced,0.26523733139038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,16,power_law_1.2,2.47326717376709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,16,1,balanced,0.26687467098236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,16,1,balanced,0.2738506595293681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,16,1,balanced,0.27875200907389325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,16,1,balanced,0.0773119976123174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,16,power_law_1.2,2.3031808853149416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,16,1,balanced,0.28306132555007935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,16,1,balanced,0.29097600777943927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,16,1,balanced,0.3485706647237142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,16,1,balanced,0.3669653336207072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,16,power_law_1.2,0.07466880083084107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,16,1,balanced,0.40250666936238605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,16,power_law_1.2,0.09700480103492737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,16,1,balanced,0.4376266797383626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,16,power_law_1.2,3.5766975402832033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,16,power_law_1.2,0.08544639945030212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,16,1,balanced,0.6858186721801758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,16,power_law_1.2,3.242950439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,16,power_law_1.2,0.08414080142974853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,16,1,balanced,0.7590719858805338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,16,power_law_1.2,0.17681280374526978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,16,1,balanced,1.0533226331075032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,16,power_law_1.2,0.15941120386123658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,16,1,balanced,0.12155200044314067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,16,1,balanced,0.14477333426475525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,16,1,balanced,1.3891572952270508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,16,1,balanced,0.1750026742617289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,16,power_law_1.2,0.1328511953353882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,16,1,balanced,0.17655466000239053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,16,1,balanced,0.17742933829625449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,16,1,balanced,0.17779199282328287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,16,power_law_1.2,0.10574079751968384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,16,1,balanced,0.19806933403015137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,16,1,balanced,1.8796480496724446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,16,1,balanced,0.19792532920837402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,16,power_law_1.2,0.10120960474014282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,16,1,balanced,0.20295466979344687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,16,1,balanced,0.20746133724848428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,16,1,balanced,0.2114773392677307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,16,power_law_1.2,0.11679359674453735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,16,1,balanced,0.21741332610448202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,16,1,balanced,0.22971200942993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,16,power_law_1.2,0.12749439477920532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,16,1,balanced,2.52622397740682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,16,power_law_1.2,4.354931259155274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,16,power_law_1.2,0.16002559661865234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,16,1,power_law_1.01,0.035411199927330016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,16,1,power_law_1.01,0.03199360072612763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,16,1,power_law_1.01,0.04779520034790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,16,power_law_1.2,2.590713691711426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,16,1,power_law_1.01,0.03491199910640717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,16,1,power_law_1.01,0.06901119947433472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,16,1,power_law_1.01,0.05186560153961182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,16,1,power_law_1.01,0.10118399858474732
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,16,power_law_1.2,0.19225599765777587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,16,1,power_law_1.01,0.07220479846000671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,16,1,power_law_1.01,0.134278404712677
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,16,power_law_1.2,0.26893439292907717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,16,1,power_law_1.01,0.08710399866104127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,16,power_law_1.2,5.259436798095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,16,1,power_law_1.01,0.118777596950531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,16,power_law_1.2,0.34568960666656495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,16,1,power_law_1.2,1.5375167846679687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,16,1,power_law_1.01,0.15468159914016724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,16,1,balanced,0.2890666723251343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,16,1,balanced,0.3083146611849467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,16,1,power_law_1.2,0.03551360070705414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,16,1,power_law_1.01,0.15967999696731566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,16,1,balanced,0.34694933891296387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,16,1,power_law_1.2,0.04886400103569031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,16,1,balanced,0.38385601838429767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,16,1,power_law_1.2,0.06974080204963684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,16,1,balanced,0.5937973260879517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,16,1,power_law_1.2,0.093094402551651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,16,1,balanced,0.66866135597229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,16,power_law_1.2,0.4689343929290771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,16,1,power_law_1.2,0.12725119590759276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,16,1,balanced,0.9336000283559164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,16,1,power_law_1.2,0.21647999286651612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,16,1,power_law_1.01,0.21665279865264891
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,16,power_law_1.2,0.6220863819122314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,16,1,balanced,1.2226666609446208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,16,1,power_law_1.2,0.24045441150665284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,16,1,power_law_1.01,0.24155519008636475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,16,1,power_law_1.2,0.2403968095779419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,16,1,power_law_1.01,0.24232959747314453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,16,1,balanced,1.6655839284261067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,16,1,power_law_1.2,0.18988159894943238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,16,1,power_law_1.01,0.19529600143432618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,16,1,power_law_1.01,0.15871360301971435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,16,1,power_law_1.2,0.19779839515686035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,16,1,power_law_1.01,0.19338879585266114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,16,1,power_law_1.01,0.16329599618911744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,16,1,power_law_1.01,0.20965759754180907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,16,1,power_law_1.2,0.24897921085357666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,16,1,balanced,2.218261400858561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,16,1,power_law_1.01,0.1686911940574646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,16,1,power_law_1.01,0.22200319766998292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,16,1,power_law_1.2,0.23180799484252929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,16,1,power_law_1.01,0.17783039808273315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,16,1,power_law_1.2,0.02913280129432678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,16,1,power_law_1.01,0.23432319164276122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,16,1,power_law_1.2,0.25112318992614746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,16,1,power_law_1.01,0.1764672040939331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,16,1,power_law_1.2,0.039743998646736146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,16,1,power_law_1.01,0.24846720695495605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,16,1,power_law_1.01,0.20248959064483643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,16,1,power_law_1.2,0.2671999931335449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,16,1,power_law_1.2,0.049491199851036075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,16,1,power_law_1.01,0.2718463897705078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,16,1,power_law_1.01,0.20889599323272706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,16,1,power_law_1.2,0.27587199211120605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,16,1,power_law_1.01,0.27981441020965575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,16,1,power_law_1.01,0.21722240447998048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,16,1,power_law_1.2,0.28022398948669436
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,16,power_law_1.2,0.9862208366394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,16,1,power_law_1.01,0.2923327922821045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,16,1,power_law_1.01,0.23188478946685792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,16,1,power_law_1.2,0.297107195854187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,16,1,power_law_1.01,0.3546175956726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,16,1,power_law_1.01,0.29717121124267576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,16,1,power_law_1.2,0.3576256036758423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,16,1,power_law_1.01,0.37838718891143797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,16,1,power_law_1.01,0.32247679233551024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,16,1,power_law_1.2,0.3966207981109619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,16,1,power_law_1.01,0.47631359100341797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,16,1,power_law_1.01,0.3975744009017944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,16,1,power_law_1.2,0.483846378326416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,16,1,power_law_1.01,0.5307007789611816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,16,1,power_law_1.2,1.4546943664550782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,16,1,power_law_1.2,0.0654528021812439
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,16,power_law_1.2,1.3342720031738282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,16,balanced,0.05143466591835022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,16,balanced,0.051728000243504844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,16,balanced,0.041834667325019836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,16,1,power_law_1.2,0.07769600152969361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,16,balanced,0.06557866434256236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,16,1,power_law_1.01,0.6794112205505372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,16,balanced,0.04173333446184794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,16,balanced,0.09063466389973958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,16,balanced,0.05203199883302053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,16,balanced,0.16173332929611206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,16,1,power_law_1.2,0.11574399471282959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,16,balanced,0.2704319953918457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,16,power_law_1.2,7.547353363037109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,16,1,power_law_1.2,0.15639679431915282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,16,balanced,0.38433067003885907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,16,balanced,0.38787734508514404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,16,1,power_law_1.2,0.1604032039642334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,16,balanced,0.44013333320617676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,16,1,power_law_1.2,0.16794240474700928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,16,balanced,0.4420586824417114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,16,balanced,0.05825600028038025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,16,balanced,0.06031466523806254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,16,balanced,0.07452799876530965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,16,balanced,0.08922132849693298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,16,balanced,0.10737066467603047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,16,balanced,0.16325333714485168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,16,1,power_law_1.01,0.4546048164367676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,16,balanced,0.24786667029062906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,16,balanced,0.2542720039685567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,16,balanced,0.4650719960530599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,16,balanced,0.47009599208831787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,16,balanced,0.0682773341735204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,16,balanced,0.09834667046864827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,16,balanced,0.476858655611674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,16,1,power_law_1.2,0.5396287918090821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,16,balanced,0.16562133034070334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,16,1,power_law_1.01,0.8020031929016114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,16,balanced,0.4844319820404053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,16,balanced,0.2529866695404053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,16,balanced,0.2550879915555318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,16,balanced,0.49326932430267334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,16,balanced,0.25245867172876996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,16,1,power_law_1.2,0.6860288143157959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,16,balanced,0.26683733860651654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,16,power_law_1.2,3.339904022216797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,16,balanced,0.35913066069285077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,16,1,power_law_1.01,1.0937919616699219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,16,balanced,0.49287466208140057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,16,balanced,0.3696746826171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,16,1,power_law_1.2,0.8075200080871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,16,balanced,0.37652266025543213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,16,balanced,0.5055786768595377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,16,balanced,0.3877386649449666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,16,balanced,0.5117599964141846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,16,balanced,0.39956267674763996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,16,1,power_law_1.01,1.3827775955200194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,16,balanced,0.5225919882456461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,16,balanced,0.41389866669972736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,16,1,power_law_1.2,1.1071423530578612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,16,balanced,0.5338879823684692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,16,1,power_law_1.01,0.5852543830871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,16,balanced,0.4333440065383911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,16,balanced,0.5439306497573853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,16,balanced,0.4941546519597371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,16,balanced,0.5675413211186727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,16,balanced,0.5294026533762614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,16,1,power_law_1.01,0.7004735946655274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,16,1,power_law_1.01,1.9396352767944336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,16,balanced,0.6469013293584188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,16,balanced,0.5064160029093424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,16,balanced,0.6769546667734782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,16,1,power_law_1.01,0.9574272155761718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,16,balanced,0.5233866771062216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,16,1,power_law_1.2,1.3897664070129394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,16,1,power_law_1.2,0.1698240041732788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,16,balanced,0.7412532965342203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,16,balanced,0.5369600057601929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,16,1,power_law_1.2,0.1819584012031555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,16,balanced,0.5650666554768881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,16,1,power_law_1.01,2.5397567749023438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,16,1,power_law_1.2,0.19286400079727173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,16,1,power_law_1.01,1.2166912078857421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,16,balanced,0.6153440078099569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,16,1,power_law_1.2,0.18672000169754027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,16,balanced,0.6667359670003256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,16,power_law_1.01,0.1512063980102539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,16,1,power_law_1.2,0.2038975954055786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,16,balanced,0.7703893184661865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,16,power_law_1.01,0.24674561023712158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,16,1,power_law_1.2,0.21137919425964355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,16,1,power_law_1.01,1.7216320037841797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,16,balanced,0.59661332766215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,16,1,power_law_1.2,0.21766400337219238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,16,power_law_1.01,0.3356096029281616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,16,balanced,0.6644853353500366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,16,1,power_law_1.2,0.23422720432281494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,16,balanced,0.8025866349538168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,16,power_law_1.01,0.3346560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,16,1,power_law_1.2,0.302239990234375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,16,balanced,0.9374346733093262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,16,balanced,0.8009226322174072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,16,1,power_law_1.01,2.2472448348999023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,16,1,power_law_1.2,0.32667520046234133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,16,power_law_1.01,0.34344959259033203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,16,balanced,1.2123146851857503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,16,balanced,0.9273013273874918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,16,1,power_law_1.2,0.4045567989349365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,16,power_law_1.01,0.08814079761505127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,16,power_law_1.01,0.3543551921844482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,16,balanced,1.056389331817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,16,1,power_law_1.2,0.45876479148864746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,16,power_law_1.01,0.14915839433670045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,16,balanced,1.5926879247029622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,16,power_law_1.01,0.3354880094528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,16,balanced,0.8705600102742513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,16,1,power_law_1.2,0.5915775775909424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,16,balanced,0.031290667752424874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,16,power_law_1.01,0.20428800582885742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,16,balanced,0.033344000577926636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,16,balanced,0.03349866718053818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,16,balanced,2.1415039698282876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,16,balanced,0.039813332259655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,16,power_law_1.01,0.33705599308013917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,16,balanced,0.06306133170922597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,16,1,power_law_1.2,0.7064064025878907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,16,power_law_1.01,0.20122880935668946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,16,balanced,0.09267733494440715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,16,balanced,0.124917338291804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,16,balanced,0.12553600470225015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,16,balanced,0.12627733747164407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,16,power_law_1.01,0.20602240562438964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,16,balanced,0.12644267082214355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,16,balanced,0.12920000155766806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,16,balanced,0.1316373348236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,16,power_law_1.01,0.21674880981445313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,16,balanced,0.1346879998842875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,16,balanced,0.1790293256441752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,16,balanced,0.17297067244847616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,16,power_law_1.01,0.22071681022644044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,16,balanced,0.16236266493797302
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,16,balanced,0.35229865709940594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,16,balanced,0.2603360017140706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,16,power_law_1.01,0.2187648057937622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,16,1,power_law_1.2,0.9679424285888671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,16,balanced,0.18639999628067017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,16,balanced,1.0755093097686768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,16,balanced,0.19732799132665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,16,power_law_1.01,0.21824638843536376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,16,balanced,0.2158720095952352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,16,power_law_1.01,0.3917248010635376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,16,balanced,1.491653283437093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,16,balanced,0.2569546699523926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,16,1,power_law_1.2,1.228108787536621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,16,power_law_1.01,0.22232320308685302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,16,balanced,2.8014612197875977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,16,balanced,0.2998453378677368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,16,power_law_1.01,0.35822720527648927
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,16,balanced,0.38446398576100665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,16,power_law_1.01,0.3289407968521118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,16,balanced,1.8975946108500164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,16,1,power_law_1.2,1.9690176010131837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,16,balanced,0.5112799803415934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,16,power_law_1.01,0.4266047954559326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,16,power_law_1.01,0.3308480024337769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,16,balanced,1.306554635365804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,16,power_law_1.01,0.09903360009193421
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,16,balanced,0.6684266726175944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,16,power_law_1.01,0.4640960216522217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,16,balanced,2.5184106826782227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,16,power_law_1.01,0.34013440608978274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,16,power_law_1.01,0.15471359491348266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,16,balanced,0.945904016494751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,16,power_law_1.01,0.44725761413574217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,16,power_law_1.01,0.3583103895187378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,16,power_law_1.01,0.24057600498199463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,16,power_law_1.01,0.47475199699401854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,16,power_law_1.01,0.3726912021636963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,16,power_law_1.01,0.23327360153198243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,16,balanced,3.543621381123861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,16,power_law_1.01,0.47191681861877444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,16,power_law_1.01,0.37925760746002196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,16,power_law_1.01,0.24222080707550048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,16,power_law_1.01,0.4029376029968262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,16,1,power_law_1.2,1.7441984176635743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,16,power_law_1.01,0.4817920207977295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,16,balanced,1.7654132843017578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,16,balanced,4.775472005208333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,16,power_law_1.01,0.5127999782562256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,16,power_law_1.2,0.1530751943588257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,16,power_law_1.01,0.6024511814117431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,16,balanced,2.2912000020345054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,16,power_law_1.2,0.24702720642089843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,16,power_law_1.01,0.6961023807525635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,16,power_law_1.2,0.31182079315185546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,16,power_law_1.01,0.4972095966339111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,16,balanced,3.01802666982015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,16,power_law_1.01,0.24651520252227782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,16,power_law_1.01,0.8586560249328613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,16,power_law_1.2,0.33717761039733884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,16,power_law_1.01,0.5167103767395019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,16,power_law_1.2,0.08851839900016785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,16,power_law_1.01,0.19254399538040162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,16,power_law_1.2,0.34495360851287843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,16,power_law_1.01,1.029139232635498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,16,power_law_1.01,0.611955213546753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,16,power_law_1.2,0.15029759407043458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,16,power_law_1.01,0.2380608081817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,16,power_law_1.2,0.19102079868316652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,16,power_law_1.2,0.3569279909133911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,16,power_law_1.01,0.6516863822937011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,16,power_law_1.01,0.42609281539916993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,16,power_law_1.01,1.3599488258361816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,16,power_law_1.2,0.20038399696350098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,16,power_law_1.01,0.7433536052703857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,16,power_law_1.01,0.4440512180328369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,16,power_law_1.2,0.20622720718383789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,16,power_law_1.01,0.8351743698120118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,16,power_law_1.01,1.6816640853881837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,16,power_law_1.01,0.4169663906097412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,16,power_law_1.2,0.1971392035484314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,16,power_law_1.01,0.057030397653579715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,16,power_law_1.01,0.4476160049438477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,16,power_law_1.01,1.0245759963989258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,16,power_law_1.2,0.19493119716644286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,16,power_law_1.01,0.08453119993209839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,16,power_law_1.2,0.22783360481262208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,16,power_law_1.01,2.364575958251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,16,power_law_1.01,0.11159039735794067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,16,power_law_1.01,1.2523391723632813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,16,power_law_1.2,0.2048703908920288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,16,power_law_1.2,0.38309760093688966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,16,power_law_1.01,0.1076159954071045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,16,power_law_1.2,0.20531840324401857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,16,1,power_law_1.2,2.2693759918212892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,16,power_law_1.01,1.6265087127685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,16,power_law_1.2,0.3404223918914795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,16,power_law_1.2,0.325926399230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,16,power_law_1.01,3.0743871688842774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,16,power_law_1.2,0.36706559658050536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,16,power_law_1.2,0.09827839732170104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,16,power_law_1.2,0.33962240219116213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,16,power_law_1.01,1.9585023880004884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,16,power_law_1.2,0.3768831968307495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,16,power_law_1.01,0.45733118057250977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,16,power_law_1.2,0.15631359815597534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,16,power_law_1.2,0.36338560581207274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,16,power_law_1.2,0.45098237991333007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,16,power_law_1.2,0.21484799385070802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,16,power_law_1.01,0.48574080467224123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,16,power_law_1.2,0.3686336040496826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,16,1,power_law_1.2,2.544281578063965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,16,power_law_1.2,0.241811203956604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,16,power_law_1.2,0.4324031829833984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,16,1,balanced,0.039061332742373146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,16,power_law_1.01,0.114028799533844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,16,power_law_1.01,0.49126400947570803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,16,power_law_1.2,0.2346560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,16,power_law_1.2,0.45871357917785643
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,16,power_law_1.01,0.11620479822158813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,16,power_law_1.01,0.5145792007446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,16,power_law_1.2,0.24659841060638427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,16,power_law_1.2,0.4701119899749756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,16,power_law_1.01,0.11330560445785523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,16,power_law_1.01,0.5410560131072998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,16,power_law_1.2,0.18940800428390503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,16,power_law_1.2,0.4767615795135498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,16,power_law_1.01,0.10734080076217652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,16,power_law_1.01,0.610694408416748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,16,power_law_1.2,0.22388479709625245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,16,power_law_1.2,0.5053760051727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,16,power_law_1.01,0.11639679670333862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,16,power_law_1.01,0.7105855941772461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,16,power_law_1.2,0.3741568088531494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,16,power_law_1.2,0.5104191780090332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,16,1,balanced,0.05100800096988678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,16,power_law_1.2,0.43985280990600584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,16,power_law_1.01,2.788025665283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,16,1,balanced,0.06843733290831248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,16,power_law_1.01,0.11859840154647827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,16,1,balanced,0.10834133625030518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,16,1,balanced,0.22033600012461343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,16,power_law_1.2,0.39461119174957277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,16,power_law_1.2,0.6251520156860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,16,1,balanced,0.39105598131815594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,16,power_law_1.01,0.86428804397583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,16,power_law_1.01,0.10858240127563476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,16,1,balanced,0.5214720169703165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,16,power_law_1.2,0.4186560153961182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,16,1,balanced,0.5220586856206259
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,16,power_law_1.01,0.11325440406799317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,16,power_law_1.2,0.6693120002746582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,16,1,balanced,0.4742933511734009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,16,power_law_1.01,0.9943424224853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,16,power_law_1.2,0.4697408199310303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,16,1,balanced,0.4772640069325765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,16,power_law_1.01,0.1173248052597046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,16,1,balanced,0.4907253185908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,16,power_law_1.2,0.7778304100036622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,16,power_law_1.2,0.5295167922973633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,16,power_law_1.01,3.799923324584961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,16,1,balanced,0.4878400166829427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,16,power_law_1.01,0.14199680089950562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,16,1,balanced,0.03172266731659571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,16,1,balanced,0.49218134085337323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,16,1,balanced,0.03551466763019562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,16,1,balanced,0.05020266771316528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,16,power_law_1.2,0.6135744094848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,16,1,balanced,0.49619734287261963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,16,power_law_1.01,0.14573440551757813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,16,1,balanced,0.5040320158004761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,16,1,balanced,0.5097066561381022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,16,power_law_1.2,0.7049920082092285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,16,power_law_1.01,0.1331007957458496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,16,power_law_1.2,0.42560639381408694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,16,1,balanced,0.5228319962819418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,16,power_law_1.01,0.2300031900405884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,16,1,balanced,0.7003093560536703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,16,power_law_1.2,0.41867518424987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,16,power_law_1.2,0.8852864265441894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,16,1,balanced,0.7181706428527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,16,power_law_1.01,0.16787840127944947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,16,1,balanced,0.7561919689178467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,16,power_law_1.2,0.41157121658325196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,16,power_law_1.2,1.0484479904174804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,16,power_law_1.01,0.17175040245056153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,16,1,balanced,0.7964320182800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,16,power_law_1.2,0.4352384090423584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,16,power_law_1.01,1.3666303634643555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,16,1,balanced,0.8770666917165121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,16,power_law_1.01,0.20095999240875245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,16,power_law_1.2,0.8766847610473633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,16,power_law_1.2,1.419148826599121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,16,power_law_1.2,0.48351359367370605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,16,1,balanced,0.9631840387980143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,16,power_law_1.01,0.22024960517883302
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,16,power_law_1.2,0.056959998607635495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,16,power_law_1.2,1.083187198638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,16,1,balanced,1.1331573327382405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,16,power_law_1.2,0.48316159248352053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,16,power_law_1.01,0.2750720024108887
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,16,power_law_1.2,0.08346239924430847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,16,power_law_1.2,0.5114687919616699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,16,power_law_1.2,1.7587903976440429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,16,1,balanced,1.6628212928771973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,16,power_law_1.01,0.33949439525604247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,16,power_law_1.2,0.10695680379867553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,16,power_law_1.2,1.3078207969665527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,16,power_law_1.2,0.5330560207366943
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,16,power_law_1.01,0.45795841217041017
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,16,power_law_1.2,0.10487680435180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,16,1,balanced,2.0092533429463706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,16,power_law_1.2,0.6316480159759521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,16,1,balanced,0.07179733117421468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,16,power_law_1.01,0.5723648071289062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,16,1,balanced,0.11340799927711487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,16,power_law_1.2,1.6931903839111329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,16,power_law_1.2,2.587308883666992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,16,1,balanced,0.20195732514063516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,16,1,balanced,0.33478399117787677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,16,power_law_1.2,0.7268159866333008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,16,1,balanced,0.3346986770629883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,16,1,balanced,2.7218828201293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,16,power_law_1.01,1.624870491027832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,16,1,balanced,0.3272426724433899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,16,power_law_1.01,0.8465920448303222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,16,1,balanced,0.3384213447570801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,16,1,balanced,0.36790398756663006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,16,power_law_1.2,0.9150848388671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,16,power_law_1.2,2.1282495498657226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,16,power_law_1.01,1.2582015991210938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,16,power_law_1.2,3.301375961303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,16,power_law_1.2,1.0506303787231446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,16,power_law_1.01,2.3693056106567383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,16,1,power_law_1.01,0.03826560080051422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,16,power_law_1.2,0.11367679834365844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,16,1,power_law_1.01,0.046911999583244324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,16,power_law_1.2,3.0118207931518555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,16,power_law_1.2,1.4385855674743653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,16,1,power_law_1.01,0.06725119948387145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,16,power_law_1.01,2.9496320724487304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,16,1,power_law_1.01,0.10688639879226684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,16,1,balanced,0.3694773515065511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,16,power_law_1.2,1.7952512741088866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,16,1,balanced,0.37331732114156085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,16,1,power_law_1.01,0.19328000545501708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,16,1,balanced,0.37827201684316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,16,1,balanced,0.38576531410217285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,16,1,balanced,0.3933493296305339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,16,1,power_law_1.01,0.26462719440460203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,16,1,balanced,0.40584532419840497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,16,power_law_1.2,3.9676544189453127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,16,1,balanced,0.566485325495402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,16,1,power_law_1.01,0.3094336032867432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,16,1,balanced,0.5850133498509725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,16,1,power_law_1.01,0.031142398715019226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,16,power_law_1.2,2.7123008728027345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,16,1,balanced,0.6255946556727091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,16,1,power_law_1.01,0.03434880077838898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,16,power_law_1.2,0.11094399690628051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,16,power_law_1.01,4.557478332519532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,16,1,balanced,0.6677119731903076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,16,1,balanced,0.7542400360107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,16,1,power_law_1.01,0.04640640020370483
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,16,power_law_1.2,0.10779520273208618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,16,1,balanced,0.8402933279673258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,16,1,power_law_1.01,0.06840320229530335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,16,1,balanced,1.0149226983388264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,16,power_law_1.2,0.1038591980934143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,16,power_law_1.2,3.1663232803344727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,16,1,power_law_1.01,0.10397440195083618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,16,power_law_1.2,0.1121791958808899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,16,1,power_law_1.01,0.15036159753799438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,16,power_law_1.2,0.10776319503784179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,16,1,power_law_1.01,0.19742720127105712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,16,1,power_law_1.01,0.3619647979736328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,16,power_law_1.01,5.5829822540283205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,16,power_law_1.2,0.11029759645462037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,16,1,power_law_1.01,0.2515199899673462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,16,1,power_law_1.01,0.3098687887191772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,16,power_law_1.2,0.10805120468139648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,16,1,power_law_1.01,0.23114240169525146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,16,1,power_law_1.01,0.3172032117843628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,16,1,balanced,0.03136000037193298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,16,power_law_1.2,4.533465576171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,16,1,balanced,0.032730666299661
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,16,power_law_1.2,0.1257472038269043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,16,1,balanced,0.03573866685231527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,16,1,balanced,0.047797332207361855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,16,1,power_law_1.01,0.23513600826263428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,16,1,power_law_1.01,0.423737621307373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,16,power_law_1.2,0.13279999494552613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,16,1,balanced,1.4727786382039387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,16,1,power_law_1.01,0.32238719463348386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,16,1,power_law_1.01,0.4270463943481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,16,power_law_1.2,0.13266559839248657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,16,1,balanced,1.8166507085164387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,16,1,power_law_1.01,0.43289599418640134
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,16,power_law_1.2,0.1363711953163147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,16,1,power_law_1.01,0.4430208206176758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,16,power_law_1.2,0.2284480094909668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,16,1,balanced,2.4431039492289224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,16,1,power_law_1.01,0.4516416072845459
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,16,power_law_1.2,0.21006720066070556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,16,power_law_1.2,6.339315032958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,16,1,balanced,0.07437866429487865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,16,1,power_law_1.01,0.4626880168914795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,16,power_law_1.2,0.1709887981414795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,16,1,balanced,0.11635733644167583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,16,1,balanced,0.16088533401489258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,16,1,power_law_1.2,0.03749760091304779
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,16,1,balanced,0.16335999965667725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,16,1,balanced,0.16542933384577432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,16,power_law_1.2,0.20429439544677735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,16,1,power_law_1.01,0.47587199211120607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,16,1,balanced,0.16776533921559653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,16,1,power_law_1.2,0.04583680033683777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,16,1,balanced,0.1722666621208191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,16,1,power_law_1.01,0.3263360023498535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,16,1,balanced,0.17704532543818155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,16,power_law_1.2,0.22950398921966553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,16,1,power_law_1.2,0.06616960167884826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,16,1,balanced,0.18393067518870035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,16,1,power_law_1.01,0.6542975902557373
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,16,1,balanced,0.19301333030064902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,16,1,balanced,0.2032853364944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,16,1,power_law_1.01,0.3359296083450317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,16,1,power_law_1.2,0.10633599758148193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,16,power_law_1.2,0.2995199918746948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,16,1,balanced,0.212826669216156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,16,1,balanced,0.25827733675638836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,16,1,power_law_1.01,0.6756735801696777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,16,1,balanced,0.2999253273010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,16,1,power_law_1.2,0.1661695957183838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,16,1,power_law_1.01,0.34172160625457765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,16,power_law_1.2,0.35438721179962157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,16,1,power_law_1.01,0.7321023941040039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,16,1,power_law_1.01,0.3559295892715454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,16,1,power_law_1.2,0.24947841167449952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,16,power_law_1.2,0.47997441291809084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,16,1,power_law_1.01,0.3612031936645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,16,1,power_law_1.01,0.7871103763580323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,16,1,power_law_1.2,0.29022719860076907
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,16,power_law_1.2,0.5997439861297608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,16,1,power_law_1.01,0.3801408052444458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,16,1,power_law_1.2,0.35990400314331056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,16,1,power_law_1.01,0.9590656280517578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,16,1,power_law_1.01,0.526310396194458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,16,1,power_law_1.2,0.30334720611572263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,16,power_law_1.2,0.8886912345886231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,16,1,power_law_1.01,1.0725631713867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,16,1,power_law_1.01,0.5492159843444824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,16,1,balanced,0.2869973381360372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,16,1,power_law_1.2,0.3066688060760498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,16,1,balanced,0.3429386615753174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,16,1,balanced,0.35620800654093426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,16,1,power_law_1.01,0.6113088130950928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,16,1,power_law_1.2,0.42035841941833496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,16,1,power_law_1.01,1.357363224029541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,16,1,balanced,0.4392746686935425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,16,power_law_1.2,1.3945856094360352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,16,1,balanced,0.6004746754964193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,16,1,power_law_1.01,0.6662911891937255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,16,1,power_law_1.2,0.4212800025939941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,16,1,power_law_1.2,0.03047040104866028
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,16,1,balanced,0.749509334564209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,16,1,power_law_1.2,0.4302015781402588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,16,1,power_law_1.2,0.03404799997806549
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,16,1,balanced,0.9362666606903076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,16,1,power_law_1.2,0.04845440089702606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,16,1,power_law_1.01,0.027513599395751952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,16,1,balanced,0.9797600110371908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,16,1,power_law_1.2,0.06812800168991089
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,16,1,power_law_1.01,0.028883200883865357
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,16,1,balanced,1.365850607554118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,16,1,power_law_1.2,0.09902080297470092
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,16,1,power_law_1.01,0.03206399977207184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,16,1,power_law_1.2,0.1371072053909302
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,16,1,power_law_1.01,0.0470335990190506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,16,1,power_law_1.01,1.6464832305908204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,16,1,power_law_1.2,0.18995200395584105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,16,1,power_law_1.01,0.06727679967880248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,16,1,power_law_1.2,0.4406720161437988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,16,1,power_law_1.01,0.0955136001110077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,16,1,power_law_1.2,0.24606719017028808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,16,1,power_law_1.2,0.45160322189331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,16,1,power_law_1.01,0.11839359998703003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,16,1,power_law_1.2,0.22522239685058593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,16,1,power_law_1.01,2.2006336212158204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,16,1,power_law_1.2,0.4642752170562744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,16,1,power_law_1.01,0.13306879997253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,16,1,power_law_1.2,0.23198719024658204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,16,1,power_law_1.2,0.4825984001159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,16,1,power_law_1.01,0.13806719779968263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,16,1,power_law_1.2,0.3203007936477661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,16,1,power_law_1.01,0.8105983734130859
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,16,1,power_law_1.01,0.14343039989471434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,16,1,power_law_1.2,0.6534848213195801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,16,1,power_law_1.01,2.77640323638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,16,1,power_law_1.2,0.3265599966049194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,16,1,power_law_1.01,0.15231360197067262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,16,1,power_law_1.2,0.681766414642334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,16,1,power_law_1.01,0.9316672325134278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,16,1,power_law_1.2,0.3305984020233154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,16,1,power_law_1.01,0.1592128038406372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,16,1,power_law_1.2,0.746937608718872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,16,1,power_law_1.01,1.1966591835021974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,16,1,power_law_1.2,0.3400511980056763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,16,1,power_law_1.01,0.17015680074691772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,16,1,power_law_1.2,0.8096447944641113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,16,1,power_law_1.2,0.35190401077270506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,16,1,power_law_1.01,0.18368639945983886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,16,1,power_law_1.01,1.4617919921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,16,1,power_law_1.01,0.19672319889068604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,16,1,power_law_1.2,0.9829312324523926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,16,1,power_law_1.2,0.3615936040878296
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,16,1,power_law_1.01,0.211296010017395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,16,1,power_law_1.2,0.37621119022369387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,16,1,power_law_1.2,1.1021568298339843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,16,1,power_law_1.01,0.24058880805969238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,16,1,power_law_1.01,1.9810176849365235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,16,1,power_law_1.2,0.5356544017791748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,16,1,power_law_1.01,0.30636799335479736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,16,1,power_law_1.2,0.5637951850891113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,16,1,power_law_1.2,1.3932864189147949
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,16,1,power_law_1.01,0.2742271900177002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,16,1,power_law_1.2,0.6220928192138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,16,1,power_law_1.01,0.3569535970687866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,16,1,power_law_1.01,2.5199167251586916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,balanced,0.033413333197434746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,16,1,power_law_1.2,1.6742464065551759
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,16,1,power_law_1.01,0.349126410484314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,16,1,power_law_1.2,0.6879551887512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,16,1,power_law_1.01,0.46039681434631347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,16,1,power_law_1.2,0.8323007583618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,16,1,power_law_1.01,0.5247615814208985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,16,1,power_law_1.2,2.252947235107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,16,1,power_law_1.2,0.9508799552917481
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,16,1,power_law_1.01,0.701632022857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,16,1,power_law_1.2,1.2213760375976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,16,1,power_law_1.2,0.027596798539161683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,16,1,power_law_1.2,2.8159936904907226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,16,1,power_law_1.2,0.029158401489257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,balanced,0.029829333225886028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,balanced,0.03155199935038885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,balanced,0.03156800071398417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,balanced,0.037658666570981346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,balanced,0.05171733101209005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,16,1,power_law_1.2,0.033369600772857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,balanced,0.0745600014925003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,balanced,0.0867199997107188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,balanced,0.08674133817354839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,balanced,0.09331732988357544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,balanced,0.0949173370997111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,16,1,power_law_1.2,0.047225600481033324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,balanced,0.12403733531634013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,balanced,0.1318933367729187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,balanced,0.1341759959856669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,balanced,0.14070933063824972
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,16,1,power_law_1.2,0.06570879817008972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,balanced,0.14798399806022644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,balanced,0.152730663617452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,balanced,0.16673066218694052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,16,1,power_law_1.2,0.09101439714431762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,balanced,0.03339199970165888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,balanced,0.03340800106525421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,balanced,0.047338664531707764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,balanced,0.06601066887378693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,balanced,0.11125333110491435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,balanced,0.11956800023714702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,16,1,power_law_1.2,0.11541119813919068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,balanced,0.12107200423876445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,balanced,0.11182933052380879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,balanced,0.11318399508794148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,balanced,0.14729066689809164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,16,1,power_law_1.2,1.48919677734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,16,1,power_law_1.2,0.13207679986953735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,16,1,power_law_1.2,0.1373311996459961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,16,1,power_law_1.2,0.14202239513397216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,16,1,power_law_1.2,2.0244352340698244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,16,1,power_law_1.2,0.04688639938831329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,balanced,0.19326933224995932
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,16,1,power_law_1.01,0.7590976238250733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,16,1,power_law_1.2,0.15070079565048217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,16,1,power_law_1.2,0.1576192021369934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,16,1,power_law_1.01,1.0392767906188964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,balanced,0.15409066279729208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,16,1,power_law_1.2,0.1702015995979309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,balanced,0.15622933705647787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,16,1,power_law_1.2,2.545523262023926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,balanced,0.16221867005030313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,balanced,0.16876266400019327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,balanced,0.042021334171295166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,balanced,0.17459199825922647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,16,1,power_law_1.2,0.18405760526657106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,balanced,0.04171733558177948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,balanced,0.18435200055440268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,balanced,0.04385066529115041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,balanced,0.04938133557637533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,balanced,0.21836799383163452
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,16,1,power_law_1.01,1.4097408294677733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,balanced,0.060080001751581825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,balanced,0.2353066603342692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,balanced,0.08077866832415263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,16,1,power_law_1.2,0.1987264037132263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,balanced,0.07853333155314128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,balanced,0.2704906662305196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,balanced,0.08057066798210144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,balanced,0.30498133103052777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,balanced,0.1535413364569346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,balanced,0.1528320014476776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,balanced,0.37384533882141113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,balanced,0.1567626694838206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,balanced,0.2128373384475708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,balanced,0.16249066591262817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,balanced,0.44047999382019043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,balanced,0.2505813241004944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,balanced,0.16516799728075662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,balanced,0.17481066783269247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,balanced,0.6277173360188802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,balanced,0.18090667327245077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,balanced,0.18922134240468344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,balanced,0.7653760115305582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,balanced,0.20593067010243735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,balanced,0.23637866973876953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,balanced,0.2671519915262858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,balanced,1.0923360188802083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,balanced,0.324618657430013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,balanced,0.38492798805236816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,balanced,0.5499039888381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,balanced,0.6670080025990804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,16,1,power_law_1.2,0.21493120193481446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,balanced,0.9509493509928385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,16,1,power_law_1.2,0.2421504020690918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,balanced,0.29151467482248944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,balanced,1.2336266835530598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,balanced,0.025834667185942333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,balanced,0.36897599697113037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,balanced,0.0262719988822937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,16,1,power_law_1.2,0.31228160858154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,balanced,0.4458293517430623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,balanced,0.6288266579310099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,16,1,power_law_1.2,0.2816256046295166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,balanced,0.7818559805552164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,16,1,power_law_1.2,0.36193280220031737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,balanced,1.411904017130534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,balanced,1.1211946805318196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,power_law_1.01,0.05988479852676391
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,16,1,power_law_1.2,0.3549567937850952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,balanced,1.4561866124471028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,power_law_1.01,0.09626880288124084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,16,1,power_law_1.2,0.46788477897644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,power_law_1.01,0.09632639884948731
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,16,1,power_law_1.2,0.5287295818328858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,balanced,0.02739199995994568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,power_law_1.01,0.09707520008087159
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,balanced,0.027488000690937042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,balanced,0.03164800008138021
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,balanced,0.05031466484069824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,power_law_1.01,0.09811840057373047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,16,1,power_law_1.2,0.7110847949981689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,power_law_1.01,0.09600639939308167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,power_law_1.01,0.09997439980506898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,16,1,power_law_1.2,0.7713920116424561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,power_law_1.01,0.109990394115448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,power_law_1.01,0.10488320589065551
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,16,1,power_law_1.2,1.051097583770752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,power_law_1.01,0.10851199626922607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,power_law_1.01,0.12554240226745605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,balanced,1.7953813870747883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,power_law_1.01,0.13630080223083496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,balanced,0.05202666421731313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,balanced,0.053861334919929504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,16,1,power_law_1.2,1.4208383560180664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,balanced,0.05449600021044413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,power_law_1.01,0.13201279640197755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,balanced,0.05570133527119955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,balanced,0.0545066644748052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,power_law_1.01,0.15247360467910767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,power_law_1.01,0.048416000604629514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,balanced,2.363680044809977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,power_law_1.01,0.15619839429855348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,power_law_1.01,0.0695743978023529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,power_law_1.01,0.1580415964126587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,power_law_1.01,0.05857279896736145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,power_law_1.01,0.06967679858207702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,power_law_1.01,0.1938688039779663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,power_law_1.01,0.07210239768028259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,power_law_1.01,0.06718080043792725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,power_law_1.01,0.2159424066543579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,power_law_1.01,0.07284479737281799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,power_law_1.01,0.06997119784355163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,power_law_1.01,0.23906559944152833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,power_law_1.01,0.07556480169296265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,power_law_1.01,0.06460800170898437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,power_law_1.01,0.2881792068481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,power_law_1.01,0.0753279983997345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,power_law_1.01,0.08199040293693542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,power_law_1.01,0.33334400653839114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,balanced,0.07679466903209686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,16,1,power_law_1.2,0.05093119740486145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,power_law_1.01,0.07511039972305297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,power_law_1.01,0.4256127834320068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,power_law_1.01,0.07779840230941773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,16,1,power_law_1.2,0.06198400259017944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,power_law_1.01,0.5133823871612548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,power_law_1.01,0.07808640003204345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,16,1,power_law_1.2,0.07822719812393189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,power_law_1.01,0.687718391418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,power_law_1.01,0.14788479804992677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,power_law_1.01,0.14746880531311035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,power_law_1.01,0.8751680374145507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,power_law_1.01,0.15231360197067262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,power_law_1.01,0.1561087965965271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,power_law_1.01,0.028883200883865357
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,balanced,0.07262399792671204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,power_law_1.01,0.08115199804306031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,balanced,0.06505066653092702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,power_law_1.01,0.15811840295791627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,balanced,0.13642133275667825
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,power_law_1.01,0.044300800561904906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,power_law_1.01,0.08515840172767639
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,balanced,0.14432533582051596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,power_law_1.01,0.16824320554733277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,balanced,0.10019200046857198
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,power_law_1.01,0.04344319999217987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,power_law_1.01,0.0880832016468048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,power_law_1.01,0.1761791944503784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,power_law_1.01,0.11400959491729737
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,power_law_1.01,0.04421760141849518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,power_law_1.01,0.18109439611434935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,power_law_1.01,0.12400000095367432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,power_law_1.01,0.04630399942398071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,power_law_1.01,0.20353920459747316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,power_law_1.01,0.12890880107879638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,power_law_1.01,0.2526911973953247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,power_law_1.01,0.135315203666687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,power_law_1.01,0.29240319728851316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,power_law_1.01,0.1436095952987671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,power_law_1.01,1.2909631729125977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,power_law_1.01,0.1462272047996521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,power_law_1.01,0.3700031995773315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,power_law_1.01,0.16866559982299806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,balanced,0.0724373310804367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,balanced,0.0784800002972285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,power_law_1.01,0.1918463945388794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,balanced,0.08727467060089111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,power_law_1.01,1.62109432220459
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,balanced,0.09714133540789287
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,balanced,0.11812800168991089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,power_law_1.01,0.21264638900756835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,power_law_1.2,0.0599232017993927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,power_law_1.01,0.2572223901748657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,power_law_1.2,0.09386879801750184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,power_law_1.01,0.04185599982738495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,power_law_1.01,0.3077631950378418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,power_law_1.2,0.09194880127906799
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,power_law_1.01,0.04593920111656189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,power_law_1.01,0.39823360443115235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,power_law_1.2,0.09710720181465149
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,power_law_1.01,0.043347200751304625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,power_law_1.2,0.09825279712677001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,power_law_1.01,0.4892735958099365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,power_law_1.01,0.04822399914264679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,power_law_1.01,0.4524479866027832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,power_law_1.2,0.09760000109672547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,power_law_1.01,0.05004159808158874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,power_law_1.01,0.669491195678711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,power_law_1.01,0.6125504016876221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,power_law_1.2,0.10120960474014282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,power_law_1.01,0.050783997774124144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,balanced,0.13934933145840964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,balanced,0.18939199050267538
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,power_law_1.01,0.06517120003700257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,power_law_1.01,0.7722623825073243
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,balanced,0.23390400409698486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,power_law_1.01,0.06021760106086731
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,power_law_1.01,0.058233600854873654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,power_law_1.01,1.0923839569091798
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,power_law_1.01,0.11391999721527099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,power_law_1.01,0.1
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,16,1,power_law_1.2,0.10819840431213379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,power_law_1.01,1.428934383392334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,power_law_1.01,0.07984640002250672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,16,1,power_law_1.2,0.14175360202789306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,power_law_1.01,0.07130240201950074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,power_law_1.01,0.07482240200042725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,16,1,power_law_1.2,0.16098560094833375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,power_law_1.01,2.049235153198242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,power_law_1.01,0.0895359992980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,balanced,0.3303520083427429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,16,1,power_law_1.2,0.20288000106811524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,power_law_1.01,0.10323840379714966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,power_law_1.01,0.1335039973258972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,16,1,power_law_1.2,0.2141887903213501
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,power_law_1.01,0.1569216012954712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,16,1,power_law_1.2,0.22783360481262208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,power_law_1.01,2.701862335205078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,power_law_1.01,0.2155071973800659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,power_law_1.2,0.048256000876426695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,16,1,power_law_1.2,0.23978879451751708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,power_law_1.01,0.27050879001617434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,power_law_1.2,0.06695680022239685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,power_law_1.2,0.11152640581130982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,16,1,power_law_1.2,0.2684607982635498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,power_law_1.2,0.10078719854354859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,power_law_1.2,0.07270399928092956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,power_law_1.2,0.10617599487304688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,16,1,power_law_1.2,0.2852544069290161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,power_law_1.2,0.07297279834747314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,power_law_1.2,0.12819839715957643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,power_law_1.2,0.07326080203056336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,16,1,power_law_1.2,0.2981055974960327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,power_law_1.2,0.1331007957458496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,power_law_1.2,0.07047039866447449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,power_law_1.2,0.13976320028305053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,16,1,power_law_1.2,0.3176383972167969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,power_law_1.2,0.14996479749679564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,power_law_1.2,0.07935360074043274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,16,1,power_law_1.2,0.33390719890594484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,power_law_1.2,0.15776000022888184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,power_law_1.2,0.08190079927444457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,power_law_1.01,0.8487999916076661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,power_law_1.2,0.08458880186080933
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,power_law_1.01,0.40383358001708985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,power_law_1.2,0.0832319974899292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,balanced,0.4270079930623372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,power_law_1.01,1.2274432182312012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,power_law_1.01,0.5344768047332764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,power_law_1.2,0.11351679563522339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,power_law_1.2,0.12142080068588257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,power_law_1.2,0.1229375958442688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,power_law_1.01,1.6015871047973633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,power_law_1.2,0.1352512001991272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,balanced,0.023183998962243397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,balanced,0.027232001225153606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,balanced,0.03551466763019562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,balanced,0.05421333511670431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,power_law_1.2,0.14368000030517578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,balanced,0.09443199634552002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,balanced,0.1643946667512258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,power_law_1.2,0.18200960159301757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,balanced,0.16365866859753928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,balanced,0.1644159952799479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,balanced,0.1285546620686849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,power_law_1.2,0.1942528009414673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,balanced,0.12818666299184164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,balanced,0.16510933637619019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,balanced,0.16396799683570862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,balanced,0.1646773318449656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,balanced,0.16886399189631143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,balanced,0.17085333665211996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,balanced,0.1729546586672465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,balanced,0.18075732390085855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,power_law_1.2,0.05825279951095581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,balanced,0.22354666392008463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,balanced,0.23200533787409464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,balanced,0.2535360058148702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,balanced,0.27406400442123413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,balanced,0.31142399708429974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,balanced,0.3505546649297078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,power_law_1.2,0.02898559868335724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,balanced,0.5427413384119669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,power_law_1.2,0.15242880582809448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,balanced,0.6238613526026408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,power_law_1.2,0.16664960384368896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,balanced,0.9004373550415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,power_law_1.2,0.22052481174468994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,power_law_1.2,0.19248000383377076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,power_law_1.2,0.24130558967590332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,balanced,1.1632586320241292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,balanced,0.02346666653951009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,balanced,0.02518400053183238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,balanced,0.027189334233601887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,power_law_1.2,0.3019520044326782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,power_law_1.2,0.07087360024452209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,power_law_1.2,0.3441983938217163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,power_law_1.2,0.07092480063438415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,power_law_1.2,0.4489471912384033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,power_law_1.2,0.07205119729042053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,power_law_1.2,0.04382080137729645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,power_law_1.2,0.07451519966125489
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,power_law_1.2,0.03984639942646027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,power_law_1.2,0.07292799949645996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,power_law_1.2,0.044300800561904906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,power_law_1.2,0.07415680289268493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,power_law_1.2,0.2161087989807129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,power_law_1.2,0.0426367998123169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,power_law_1.2,0.07663999795913697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,power_law_1.2,0.26880640983581544
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,power_law_1.2,0.043033599853515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,power_law_1.2,0.14295680522918702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,power_law_1.2,0.31110401153564454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,power_law_1.2,0.04869759976863861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,power_law_1.2,0.14863359928131104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,power_law_1.2,0.4114687919616699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,power_law_1.2,0.0455808013677597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,power_law_1.2,0.15160319805145264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,power_law_1.2,0.5487296104431152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,power_law_1.2,0.500819206237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,power_law_1.2,0.14949120283126832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,power_law_1.2,0.047993600368499756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,power_law_1.2,0.1585088014602661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,power_law_1.2,0.7426303863525391
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,power_law_1.2,0.047200000286102294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,power_law_1.2,0.6831103801727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,power_law_1.2,0.1711743950843811
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,power_law_1.2,0.04826880097389221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,balanced,0.03644266724586487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,power_law_1.2,0.9533568382263183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,power_law_1.2,0.8885503768920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,power_law_1.2,0.061286401748657224
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,power_law_1.2,0.0597760021686554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,power_law_1.2,0.05631359815597534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,power_law_1.2,1.2803263664245605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,power_law_1.2,1.277561569213867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,power_law_1.2,0.1151360034942627
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,power_law_1.2,0.09642239809036254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,power_law_1.2,0.0841152012348175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,power_law_1.2,1.7515775680541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,power_law_1.2,1.635513687133789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,power_law_1.2,0.073990398645401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,power_law_1.01,0.0229312002658844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,power_law_1.01,0.022015999257564544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,power_law_1.2,0.18078720569610596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,power_law_1.2,0.07934719920158387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,power_law_1.01,0.026694399118423463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,balanced,0.06069866816202799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,power_law_1.01,0.023654399812221526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,balanced,0.09706133604049683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,balanced,0.11245866616566975
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,power_law_1.2,0.0933568000793457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,balanced,0.11087999741236369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,balanced,0.1135093371073405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,power_law_1.01,0.029395198822021483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,balanced,0.11529067158699036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,power_law_1.2,0.10532480478286743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,balanced,0.13360533118247986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,balanced,0.13217066725095114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,power_law_1.01,0.037894400954246524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,balanced,0.1346773306528727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,balanced,0.13740799824396768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,balanced,0.14217600226402283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,power_law_1.01,0.051545602083206174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,balanced,0.1431893308957418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,balanced,0.1499573290348053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,balanced,0.19543999433517456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,power_law_1.01,0.06268159747123718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,balanced,0.20569600661595663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,balanced,0.22616533438364664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,balanced,0.24959999322891235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,power_law_1.01,0.09125120043754578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,balanced,0.2925013303756714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,balanced,0.33721601963043213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,balanced,0.5141173203786215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,power_law_1.2,0.18962559700012208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,balanced,0.5987360080083212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,power_law_1.01,0.0363072007894516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,power_law_1.2,0.21082239151000975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,power_law_1.01,0.05240960121154785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,balanced,0.8623519738515218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,power_law_1.2,0.2559040069580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,power_law_1.2,0.13500159978866577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,power_law_1.01,0.07824640274047852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,balanced,1.1199999650319417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,power_law_1.2,0.1615615963935852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,power_law_1.01,0.09790719747543335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,power_law_1.2,0.022649599611759184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,power_law_1.2,0.22235519886016847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,power_law_1.01,0.11793279647827148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,power_law_1.2,0.026361599564552307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,power_law_1.2,0.03612160086631775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,power_law_1.01,0.12122880220413208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,power_law_1.2,0.29091839790344237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,power_law_1.01,0.09278720021247863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,16,1,power_law_1.2,0.41836161613464357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,power_law_1.01,0.10728960037231446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,power_law_1.01,0.09499520063400269
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,power_law_1.2,0.41212158203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,power_law_1.01,0.10894080400466918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,16,1,power_law_1.2,0.4822400093078613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,16,1,balanced,0.04986133178075155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,power_law_1.01,0.13428479433059692
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,power_law_1.2,0.6074175834655762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,power_law_1.01,0.13102079629898072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,power_law_1.2,0.02247679978609085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,power_law_1.01,0.13233280181884766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,power_law_1.2,0.296998405456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,power_law_1.01,0.14577920436859132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,power_law_1.2,0.023769600689411162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,power_law_1.01,0.15859199762344361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,power_law_1.2,0.38179841041564944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,power_law_1.2,0.02871679961681366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,power_law_1.01,0.16654080152511597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,power_law_1.2,0.03775359988212586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,power_law_1.2,0.4773568153381348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,power_law_1.01,0.17238399982452393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,power_law_1.2,0.051718401908874514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,power_law_1.2,0.04782080054283142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,power_law_1.01,0.21715199947357178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,power_law_1.2,0.07007359862327575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,power_law_1.2,0.6491583824157715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,power_law_1.01,0.09673600196838379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,power_law_1.2,0.06007680296897888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,power_law_1.01,0.23267199993133544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,power_law_1.2,0.09283840060234069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,power_law_1.01,0.11227519512176513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,power_law_1.2,0.0906495988368988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,power_law_1.01,0.26467199325561525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,power_law_1.2,0.8072832107543946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,power_law_1.2,0.11431679725646973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,power_law_1.01,0.11304320096969604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,power_law_1.2,0.09203839898109437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,power_law_1.01,0.3096447944641113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,power_law_1.2,0.12277760505676269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,power_law_1.01,0.11596159934997559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,power_law_1.01,0.365721607208252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,power_law_1.2,0.1063040018081665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,power_law_1.01,0.12622720003128052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,power_law_1.01,0.43228797912597655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,power_law_1.01,0.12730879783630372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,power_law_1.01,0.1384768009185791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,power_law_1.01,0.5675327777862549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,power_law_1.01,0.14662400484085084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,power_law_1.01,0.6994368076324463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,power_law_1.01,0.18977919816970826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,power_law_1.01,0.20324480533599854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,power_law_1.01,0.974995231628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,power_law_1.2,0.09415040016174317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,power_law_1.2,1.1781375885009766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,power_law_1.2,0.09806720018386841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,power_law_1.01,1.2560576438903808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,power_law_1.2,0.11164799928665162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,16,1,balanced,0.054010664423306785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,16,1,balanced,0.06252799928188324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,power_law_1.2,0.12977919578552247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,16,1,balanced,0.08058133224646251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,power_law_1.2,1.4903167724609374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,power_law_1.2,0.13105920553207398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,power_law_1.2,0.14124799966812135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,power_law_1.2,0.15043840408325196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,power_law_1.2,0.15436160564422607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,power_law_1.01,0.2346560001373291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,power_law_1.2,2.302470397949219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,power_law_1.2,0.16603519916534423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,power_law_1.01,0.2760063886642456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,power_law_1.2,0.17650560140609742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,power_law_1.01,0.3341248035430908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,power_law_1.2,0.11047680377960205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,power_law_1.2,0.22312960624694825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,power_law_1.01,0.40090241432189944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,power_law_1.2,0.11543040275573731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,power_law_1.2,0.2370431900024414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,power_law_1.2,0.11717120409011841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,power_law_1.01,0.5333183765411377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,power_law_1.2,2.8783872604370115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,power_law_1.2,0.286899209022522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,power_law_1.2,0.12651519775390624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,power_law_1.2,0.31960320472717285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,power_law_1.01,0.6571199893951416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,16,1,power_law_1.01,0.04636160135269165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,power_law_1.2,0.3710464000701904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,power_law_1.01,0.9297727584838867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,power_law_1.2,0.44796161651611327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,power_law_1.2,0.5797440052032471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,power_law_1.01,1.1906368255615234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,power_law_1.2,0.7053567886352539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,power_law_1.2,0.1334015965461731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,power_law_1.2,0.9858176231384277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,power_law_1.2,0.13915519714355468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,power_law_1.2,0.14940160512924194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,power_law_1.2,1.2738240242004395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,power_law_1.2,0.19407999515533447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,power_law_1.2,0.20914559364318847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,power_law_1.2,0.25036799907684326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,power_law_1.2,0.28287999629974364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,power_law_1.2,0.3398591995239258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,power_law_1.2,0.4095871925354004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,power_law_1.2,0.5397759914398194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,16,1,power_law_1.2,0.5410048007965088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,power_law_1.2,0.6679872035980224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,16,1,power_law_1.2,0.6840384006500244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,power_law_1.2,0.9367615699768066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,16,1,balanced,0.12308266758918762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,power_law_1.2,1.205452823638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,16,balanced,0.03151999910672506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,16,balanced,0.0323840007185936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,16,balanced,0.033733333150545754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,16,balanced,0.035887998839219414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,16,balanced,0.04368533194065094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,16,balanced,0.055957332253456116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,16,balanced,0.06653333206971486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,16,balanced,0.06420266628265381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,16,power_law_1.01,0.05334399938583374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,16,balanced,0.06793599824110667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,16,balanced,0.06841599941253662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,16,balanced,0.07220800220966339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,16,balanced,0.07796266674995422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,16,1,power_law_1.01,0.05022720098495483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,16,balanced,0.078575998544693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,16,balanced,0.08524800340334575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,16,1,balanced,0.2052853306134542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,16,balanced,0.08990400036176045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,16,1,balanced,0.3130613366762797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,16,1,balanced,0.30806400378545123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,16,1,power_law_1.01,0.06097279787063599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,16,1,balanced,0.31307733058929443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,16,1,balanced,0.31861333052317303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,16,1,power_law_1.01,0.07764480113983155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,16,1,balanced,0.3197919925053914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,16,1,balanced,0.3250346581141154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,16,power_law_1.01,0.041145598888397215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,16,balanced,0.031498665610949196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,16,balanced,0.033370666205883026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,16,power_law_1.01,0.0686079978942871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,16,balanced,0.03772799919048945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,16,balanced,0.04789866507053375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,16,power_law_1.01,0.05136640071868896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,16,balanced,0.05781333148479462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,16,balanced,0.07260799904664357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,16,power_law_1.01,0.06635519862174988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,16,balanced,0.07240533332029979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,16,power_law_1.01,0.05052800178527832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,16,power_law_1.01,0.06322559714317322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,16,power_law_1.01,0.05064319968223572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,16,power_law_1.01,0.0523967981338501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,16,power_law_1.01,0.04679679870605469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,16,power_law_1.01,0.06968320012092591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,16,balanced,0.09539199868837993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,16,power_law_1.01,0.052934402227401735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,16,balanced,0.10521599650382996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,16,power_law_1.01,0.06885759830474854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,16,power_law_1.2,0.05242239832878113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,16,balanced,0.12980799873669943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,16,power_law_1.01,0.06481279730796814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,16,balanced,0.1443946659564972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,16,power_law_1.01,0.069760000705719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,16,power_law_1.2,0.06294400095939637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,16,balanced,0.1777013341585795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,16,power_law_1.01,0.06526079773902893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,16,power_law_1.01,0.07998080253601074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,16,power_law_1.2,0.0642304003238678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,16,power_law_1.01,0.06629120111465454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,16,power_law_1.01,0.0790336012840271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,16,power_law_1.2,0.06596480011940002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,16,power_law_1.01,0.06564480066299438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,16,power_law_1.01,0.07991679906845092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,16,power_law_1.2,0.06220800280570984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,16,power_law_1.01,0.07052159905433655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,16,power_law_1.01,0.07887359857559204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,16,power_law_1.2,0.06468480229377746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,16,balanced,0.07240533332029979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,16,power_law_1.01,0.074508798122406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,16,power_law_1.01,0.08250880241394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,16,balanced,0.08260799944400787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,16,balanced,0.0820000022649765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,16,power_law_1.2,0.06967679858207702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,16,balanced,0.08065600196520488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,16,power_law_1.01,0.07752959728240967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,16,power_law_1.01,0.0864575982093811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,16,balanced,0.08281599978605907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,16,power_law_1.2,0.07090560197830201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,16,power_law_1.01,0.08565760254859925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,16,power_law_1.2,0.07912319898605347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,16,power_law_1.01,0.0900223970413208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,16,power_law_1.2,0.07993599772453308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,16,power_law_1.01,0.09872000217437744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,16,power_law_1.2,0.08047360181808472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,16,power_law_1.01,0.11111680269241334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,16,power_law_1.2,0.07982079982757569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,16,balanced,0.2172213395436605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,16,power_law_1.01,0.13155839443206788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,16,balanced,0.30236266056696576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,16,power_law_1.2,0.08143360018730164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,16,balanced,0.368938684463501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,16,1,balanced,0.32706133524576825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,16,power_law_1.2,0.0899839997291565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,16,power_law_1.01,0.1553663969039917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,16,1,balanced,0.33081066608428955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,16,balanced,0.5179893175760905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,16,power_law_1.2,0.09416319727897644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,16,1,balanced,0.3396586577097575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,16,balanced,0.6513813336690267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,16,1,balanced,0.34538666407267254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,16,power_law_1.2,0.09987840056419373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,16,power_law_1.2,0.11546880006790161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,16,balanced,0.08473599950472514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,16,power_law_1.01,0.09178239703178406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,16,power_law_1.2,0.15327359437942506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,16,power_law_1.01,0.09804800152778625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,16,power_law_1.2,0.17352319955825807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,16,power_law_1.01,0.1154047966003418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,16,power_law_1.2,0.21948800086975098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,16,power_law_1.01,0.1423616051673889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,16,power_law_1.2,0.26093440055847167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,16,power_law_1.01,0.17209600210189818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,16,power_law_1.2,0.3727423906326294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,16,power_law_1.01,0.21235198974609376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,16,power_law_1.01,0.2560064077377319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,16,power_law_1.01,0.19466880559921265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,16,power_law_1.2,0.4546048164367676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,16,power_law_1.01,0.335097599029541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,16,power_law_1.01,0.2401344060897827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,16,balanced,0.9497653643290201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,16,balanced,0.08715732892354329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,16,power_law_1.01,0.31663360595703127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,16,power_law_1.01,0.4199103832244873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,16,balanced,0.09305066863695781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,16,balanced,0.09563199679056804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,16,balanced,0.10292266805966695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,16,balanced,1.231333335240682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,16,balanced,0.13594667116800943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,16,power_law_1.01,0.4092288017272949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,16,balanced,0.15242133537928262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,16,power_law_1.2,0.04002560079097748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,16,power_law_1.01,0.570310401916504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,16,power_law_1.01,0.7380864143371582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,16,power_law_1.01,1.0606528282165528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,16,power_law_1.2,0.6291840076446533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,16,1,power_law_1.01,0.11391999721527099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,16,power_law_1.01,1.3966400146484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,16,power_law_1.01,0.6004288196563721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,16,power_law_1.2,0.048256000876426695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,16,power_law_1.01,0.7494976043701171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,16,power_law_1.2,0.050271999835968015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,16,power_law_1.2,0.050432002544403075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,16,power_law_1.01,1.0274175643920898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,16,power_law_1.2,0.04629760086536407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,16,power_law_1.2,0.05074560046195984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,16,power_law_1.2,0.7995071887969971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,16,power_law_1.2,0.06042879819869995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,16,power_law_1.01,1.3893888473510743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,16,power_law_1.2,0.061843198537826535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,16,power_law_1.2,0.06526079773902893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,16,power_law_1.2,1.1659263610839843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,16,power_law_1.2,0.06630399823188782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,16,power_law_1.2,0.06960639953613282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,16,power_law_1.2,0.07485439777374267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,16,power_law_1.2,0.07776640057563781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,16,power_law_1.2,0.08465920090675354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,16,power_law_1.2,0.09170560240745544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,16,1,power_law_1.2,0.9162752151489257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,16,power_law_1.2,0.09904639720916748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,16,balanced,0.1830186645189921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,16,power_law_1.2,0.11149439811706544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,16,power_law_1.2,1.6119295120239259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,16,balanced,0.21279466152191162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,16,balanced,0.3026240070660909
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,16,balanced,0.3609706560770671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,16,balanced,0.5032800038655599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,16,balanced,0.6210506757100424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,16,balanced,0.9052053292592367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,16,balanced,1.1649226347605388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,16,1,power_law_1.01,0.14955519437789916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,16,1,power_law_1.01,0.17697919607162477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,16,power_law_1.2,0.13745280504226684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,16,power_law_1.2,0.15727360248565675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,16,power_law_1.2,0.19617919921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,16,power_law_1.2,0.240447998046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,16,power_law_1.2,0.33225600719451903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,16,power_law_1.2,0.4185919761657715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,16,1,power_law_1.01,0.02655999958515167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,16,1,power_law_1.01,0.03610239923000336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,16,1,power_law_1.01,0.21305599212646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,16,1,power_law_1.01,0.05185279846191406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,16,power_law_1.2,0.5798783779144288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,16,1,power_law_1.01,0.21734399795532228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,16,1,power_law_1.01,0.2344383955001831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,16,1,power_law_1.01,0.24248960018157958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,16,power_law_1.2,0.7581888198852539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,16,1,balanced,0.3314773241678874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,16,1,balanced,0.35337066650390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,16,power_law_1.2,1.0903615951538086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,16,power_law_1.2,1.4717184066772462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,16,1,balanced,0.027461332579453785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,16,1,balanced,0.03772266705830892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,16,1,balanced,0.04799466828505198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,16,1,power_law_1.01,0.2566783905029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,16,1,power_law_1.01,0.2789184093475342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,16,1,balanced,0.029296000798543293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,16,1,balanced,0.039408000806967415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,16,1,balanced,0.056218668818473816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,16,1,power_law_1.01,0.08923519849777221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,16,1,power_law_1.01,0.3022016048431396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,16,1,power_law_1.01,0.1117184042930603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,16,1,power_law_1.01,0.31943039894104003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,16,1,power_law_1.01,0.17070080041885377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,16,1,power_law_1.01,0.33386878967285155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,16,1,balanced,0.4654933214187622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,16,1,balanced,0.5133333206176758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,16,1,power_law_1.01,0.028435200452804565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,16,1,power_law_1.01,0.0377344012260437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,16,1,power_law_1.01,0.05713919997215271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,16,1,power_law_1.01,0.19592959880828859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,16,1,power_law_1.01,0.20264320373535155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,16,1,power_law_1.2,0.026464000344276428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,16,1,power_law_1.2,0.036595198512077334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,16,1,power_law_1.01,0.20911359786987305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,16,1,power_law_1.2,0.05190399885177612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,16,1,balanced,0.09682133793830872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,16,1,power_law_1.01,0.20921599864959717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,16,1,balanced,0.16920000314712524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,16,1,balanced,0.22879467407862344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,16,1,power_law_1.2,1.150476837158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,16,1,balanced,0.09114666779836018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,16,1,balanced,0.16036267081896463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,16,1,balanced,0.2160159945487976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,16,1,power_law_1.2,1.3976832389831544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,16,1,power_law_1.2,1.9683647155761719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,16,1,balanced,0.7828479607899984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,16,1,power_law_1.2,2.5428159713745115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,16,1,power_law_1.2,3.6735424041748046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,16,1,power_law_1.01,0.4147776126861572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,16,1,power_law_1.01,0.47849597930908205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,16,1,power_law_1.2,4.803936004638672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,16,1,power_law_1.01,0.0843775987625122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,16,1,power_law_1.01,0.1075711965560913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,16,1,power_law_1.01,0.16357760429382323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,16,1,power_law_1.01,0.23592960834503174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,16,1,balanced,0.8959200382232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,16,1,power_law_1.2,0.08154240250587463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,16,1,power_law_1.2,0.1053056001663208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,16,1,power_law_1.2,0.16837120056152344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,16,1,power_law_1.01,0.5281599998474121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,16,1,power_law_1.01,0.6699584007263184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,16,1,balanced,0.2384000023206075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,16,1,balanced,0.23849066098531088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,16,1,balanced,1.2986079851786296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,16,1,power_law_1.01,0.200217604637146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,16,1,balanced,0.24133867025375366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,16,1,balanced,1.7757919629414876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,16,1,power_law_1.01,0.20572800636291505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,16,1,power_law_1.01,0.21041278839111327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,16,1,power_law_1.01,0.21740798950195311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,16,1,balanced,2.522442658742269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,16,1,power_law_1.01,0.22894721031188964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,16,1,power_law_1.01,0.24286720752716065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,16,1,balanced,3.4862238566080728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,16,1,balanced,0.24251733223597208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,16,1,balanced,4.651466687520345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,16,1,power_law_1.2,0.029267200827598573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,16,1,power_law_1.2,0.038099199533462524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,16,1,power_law_1.2,0.05658239722251892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,16,balanced,0.057018667459487915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,16,balanced,0.07368533313274384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,16,balanced,0.12010666728019714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,16,1,power_law_1.01,0.20176639556884765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,16,1,power_law_1.01,0.20473599433898926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,16,1,power_law_1.01,0.2778304100036621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,16,1,power_law_1.01,0.30110719203948977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,16,1,power_law_1.01,0.36357119083404543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,16,1,power_law_1.01,0.4092671871185303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,16,1,power_law_1.01,0.5125951766967773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,16,1,power_law_1.01,0.6087679862976074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,16,1,power_law_1.01,0.832761573791504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,16,balanced,0.20523732900619507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,16,balanced,0.32660800218582153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,16,balanced,0.39760533968607586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,16,1,power_law_1.2,0.19912320375442505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,16,1,power_law_1.01,1.0369600296020507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,16,1,power_law_1.2,0.20787200927734376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,16,1,power_law_1.01,1.4572352409362792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,16,1,power_law_1.01,1.8780672073364257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,16,balanced,0.0435146689414978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,16,balanced,0.059418668349583946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,16,balanced,0.07901866734027863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,16,balanced,0.1379039982954661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,16,1,balanced,0.24422933657964072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,16,balanced,0.19226133823394775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,16,1,balanced,0.24460800488789877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,16,balanced,0.2310453255971273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,16,balanced,0.2630346616109212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,16,balanced,0.2704106569290161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,16,balanced,0.252346674601237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,16,balanced,0.2585279941558838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,16,balanced,0.4161119858423869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,16,balanced,0.4355520009994507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,16,balanced,0.45343999067942303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,16,balanced,0.4769066572189331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,16,balanced,0.4997440179189046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,16,balanced,0.5257866779963175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,16,balanced,0.5759359995524088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,16,balanced,0.39501333236694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,16,balanced,0.6539093255996704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,16,balanced,0.39895466963450116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,16,balanced,0.7344426314036051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,16,1,power_law_1.2,0.08090879917144775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,16,balanced,0.894266684850057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,16,1,power_law_1.2,0.10258560180664063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,16,balanced,1.0541653633117676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,16,1,power_law_1.2,0.16707199811935425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,16,balanced,1.4775199890136719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,16,balanced,1.795786698659261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,16,balanced,2.5470986366271973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,16,1,power_law_1.01,0.8966591835021973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,16,balanced,3.1885226567586265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,16,balanced,0.41380266348520917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,16,balanced,0.41952534516652423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,16,balanced,4.742101351420085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,16,balanced,6.1554718017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,16,balanced,0.060080001751581825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,16,balanced,0.07226666808128357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,16,balanced,0.08870933453241985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,16,balanced,0.14140266180038452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,16,balanced,0.19588265816370645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,16,balanced,0.24116800228754678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,16,balanced,0.24587732553482056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,16,balanced,0.2541439930597941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,16,balanced,0.5154079993565878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,16,balanced,0.5254400173823038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,16,balanced,0.5413066546122233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,16,balanced,0.5599093437194824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,16,balanced,0.5805279811223348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,16,balanced,0.6119199991226196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,16,balanced,0.6436266501744589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,16,balanced,0.6764159997304281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,16,1,power_law_1.2,0.2149440050125122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,16,balanced,0.7419359683990479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,16,1,power_law_1.2,0.22085120677947997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,16,balanced,0.8666986624399821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,16,balanced,0.9916319847106934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,16,balanced,1.4554293950398762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,16,balanced,1.7030720710754395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,16,1,balanced,0.7608266671498617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,16,1,balanced,0.7626506487528483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,16,balanced,2.419962724049886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,16,balanced,0.5575786828994751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,16,balanced,2.916757265726725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,16,1,power_law_1.01,1.1277631759643554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,16,1,power_law_1.01,1.3711487770080566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,16,1,power_law_1.01,1.9284927368164062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,16,1,power_law_1.2,0.20373120307922363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,16,1,power_law_1.2,0.2057919979095459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,16,1,power_law_1.01,2.492032051086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,16,balanced,4.342912038167317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,16,1,power_law_1.01,3.635232162475586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,16,balanced,5.5511519114176435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,16,1,power_law_1.01,0.6680448055267334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,16,1,power_law_1.01,4.772806549072266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,16,1,power_law_1.01,0.6875840187072754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,16,balanced,8.225151697794596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,16,1,balanced,0.26073600848515827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,16,balanced,10.639392217000326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,16,balanced,0.5777066548665365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,16,balanced,0.592298666636149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,16,power_law_1.01,0.15249919891357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,16,balanced,0.6146453221638998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,16,balanced,0.6340746482213339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,16,power_law_1.01,0.2534463882446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,16,power_law_1.01,0.20549120903015136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,16,balanced,0.6597973505655924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,16,balanced,0.7058026790618896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,16,power_law_1.01,0.2031935930252075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,16,power_law_1.01,0.3398911952972412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,16,power_law_1.01,0.21003520488739014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,16,power_law_1.01,0.3140928030014038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,16,power_law_1.01,0.21389439105987548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,16,power_law_1.01,0.2145983934402466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,16,power_law_1.01,0.24113919734954833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,16,power_law_1.01,0.24773759841918946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,16,power_law_1.01,0.23951361179351807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,16,power_law_1.01,0.2395008087158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,16,power_law_1.01,0.4087103843688965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,16,power_law_1.01,0.4146751880645752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,16,power_law_1.01,0.4402624130249023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,16,power_law_1.01,0.4617472171783447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,16,power_law_1.01,0.48867201805114746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,16,power_law_1.01,0.5182784080505372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,16,power_law_1.01,0.3433151960372925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,16,balanced,0.7910079956054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,16,power_law_1.01,0.5684607982635498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,16,balanced,0.8629333178202311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,16,power_law_1.01,0.3533888101577759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,16,power_law_1.01,0.6699711799621582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,16,balanced,1.0060266653696697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,16,power_law_1.01,0.36563839912414553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,16,power_law_1.01,0.7661888122558593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,16,balanced,1.151525338490804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,16,power_law_1.01,0.9507391929626465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,16,power_law_1.01,1.1384639739990234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,16,power_law_1.01,1.5108991622924806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,16,balanced,1.6582773526509602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,16,power_law_1.01,1.9015232086181642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,16,balanced,1.9493120511372883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,16,power_law_1.01,2.645689582824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,16,power_law_1.01,0.35393919944763186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,16,1,power_law_1.2,0.24131839275360106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,16,power_law_1.01,0.3595263957977295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,16,power_law_1.01,3.4609024047851564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,16,1,balanced,0.21906665960947672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,16,1,balanced,0.21970132986704508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,16,1,balanced,0.22311466932296753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,16,1,balanced,0.2258560061454773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,16,1,balanced,0.2265066703160604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,16,1,balanced,0.2302560011545817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,16,balanced,2.7527573903401694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,16,power_law_1.01,5.051692962646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,16,power_law_1.01,0.3547391891479492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,16,balanced,3.3683627446492515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,16,power_law_1.01,0.39854719638824465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,16,1,power_law_1.2,0.691871976852417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,16,balanced,5.033466657002767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,16,1,power_law_1.2,0.7000576019287109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,16,power_law_1.01,6.524012756347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,16,power_law_1.01,0.16643840074539185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,16,balanced,6.485904057820638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,16,power_law_1.01,0.22353920936584473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,16,power_law_1.2,0.25731840133666994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,16,power_law_1.01,0.2206655979156494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,16,power_law_1.2,0.322489595413208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,16,power_law_1.2,0.3289088010787964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,16,power_law_1.2,0.34822399616241456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,16,power_law_1.2,0.3708224058151245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,16,power_law_1.01,0.22597119808197022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,16,power_law_1.2,0.39410560131072997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,16,power_law_1.01,0.5290175914764405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,16,power_law_1.01,0.2265536069869995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,16,power_law_1.2,0.36360321044921873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,16,power_law_1.01,0.2272576093673706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,16,power_law_1.01,0.2121664047241211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,16,power_law_1.01,0.24280319213867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,16,power_law_1.01,0.49355521202087405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,16,power_law_1.01,0.49528961181640624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,16,1,power_law_1.2,0.20493440628051757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,16,power_law_1.2,0.36129279136657716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,16,power_law_1.01,0.5196352005004883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,16,1,power_law_1.2,0.20727040767669677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,16,power_law_1.2,0.37343358993530273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,16,1,power_law_1.2,0.2180351972579956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,16,power_law_1.01,0.5449408054351806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,16,1,power_law_1.2,0.22493441104888917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,16,power_law_1.2,0.3681600093841553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,16,power_law_1.01,0.5657216072082519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,16,1,power_law_1.2,0.23399679660797118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,16,power_law_1.2,0.509280014038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,16,1,power_law_1.2,0.258950400352478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,16,power_law_1.01,0.6013951778411866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,16,power_law_1.2,0.5478591918945312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,16,power_law_1.01,0.6387455940246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,16,power_law_1.2,0.5560192108154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,16,power_law_1.01,0.6892543792724609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,16,power_law_1.2,0.5980288028717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,16,power_law_1.01,0.7828735828399658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,16,power_law_1.01,0.5302720069885254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,16,power_law_1.2,0.6221759796142579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,16,power_law_1.01,0.9773759841918945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,16,power_law_1.01,0.5692351818084717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,16,power_law_1.2,0.6517695903778076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,16,power_law_1.01,0.5852287769317627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,16,power_law_1.01,1.142092800140381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,16,power_law_1.2,0.7217535972595215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,16,power_law_1.01,0.6176447868347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,16,power_law_1.2,0.8468480110168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,16,power_law_1.01,1.5378496170043945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,16,power_law_1.01,0.6481535911560059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,16,power_law_1.2,0.9364416122436523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,16,power_law_1.01,0.6967487812042237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,16,power_law_1.01,1.8519296646118164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,16,power_law_1.2,1.164358425140381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,16,power_law_1.2,1.3788928031921386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,16,power_law_1.01,2.684671974182129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,16,1,balanced,0.2694080074628194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,16,1,balanced,0.27829867601394653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,16,1,balanced,0.29582399129867554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,16,power_law_1.2,1.794905662536621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,16,1,balanced,0.3147466580073039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,16,1,balanced,0.5423466761906942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,16,1,balanced,0.5741920073827108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,16,1,balanced,0.8321279684702555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,16,power_law_1.01,3.4045185089111327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,16,power_law_1.2,2.2343040466308595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,16,1,balanced,0.9099413553873698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,16,1,balanced,1.4134292602539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,16,1,balanced,1.7556479771931965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,16,power_law_1.2,3.2783039093017576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,16,power_law_1.2,0.15234559774398804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,16,power_law_1.01,4.726291275024414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,16,power_law_1.2,0.1843840003013611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,16,power_law_1.2,0.2035968065261841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,16,1,balanced,0.9062613646189371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,16,power_law_1.2,0.20905599594116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,16,power_law_1.01,0.8096063613891602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,16,power_law_1.2,4.145817565917969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,16,power_law_1.2,0.19809919595718384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,16,power_law_1.01,0.913811206817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,16,power_law_1.2,0.2044287919998169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,16,power_law_1.01,6.2178302764892575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,16,power_law_1.01,1.1029439926147462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,16,power_law_1.2,0.23883519172668458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,16,power_law_1.2,0.22979199886322021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,16,power_law_1.01,1.346726417541504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,16,power_law_1.2,0.23767681121826173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,16,power_law_1.2,5.924601745605469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,16,power_law_1.2,0.24245760440826417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,16,power_law_1.2,0.39406719207763674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,16,power_law_1.01,1.760633659362793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,16,power_law_1.2,0.4253695964813232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,16,power_law_1.2,0.4369215965270996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,16,power_law_1.01,2.1474111557006834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,16,power_law_1.2,0.46808958053588867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,16,power_law_1.01,9.029087829589844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,16,power_law_1.2,0.4923136234283447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,16,power_law_1.2,0.5200384140014649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,16,power_law_1.2,0.5877376079559327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,16,power_law_1.2,0.6663040161132813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,16,power_law_1.2,0.7773632049560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,16,power_law_1.2,0.9782591819763183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,16,power_law_1.2,1.18090877532959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,16,1,power_law_1.01,0.8739456176757813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,16,1,power_law_1.2,0.28765439987182617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,16,power_law_1.2,1.531935977935791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,16,1,power_law_1.2,0.321343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,16,power_law_1.01,12.20074234008789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,16,1,power_law_1.2,0.37414400577545165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,16,power_law_1.2,1.9858623504638673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,16,1,power_law_1.2,0.41775999069213865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,16,power_law_1.2,8.225389099121093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,16,1,power_law_1.2,0.5279551982879639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,16,power_law_1.2,0.16791679859161376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,16,1,power_law_1.2,0.6205887794494629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,16,power_law_1.2,0.20969600677490235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,16,power_law_1.2,2.7712127685546877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,16,1,power_law_1.2,0.8409728050231934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,16,power_law_1.2,0.21415040493011475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,16,power_law_1.01,3.007814407348633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,16,power_law_1.2,0.22665600776672362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,16,1,power_law_1.2,1.0499327659606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,16,power_law_1.2,0.2050623893737793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,16,power_law_1.2,3.531398391723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,16,power_law_1.2,0.22323839664459227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,16,1,power_law_1.2,1.467686367034912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,16,power_law_1.01,3.9194305419921873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,16,power_law_1.2,0.21367039680480956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,16,power_law_1.2,0.22430078983306884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,16,power_law_1.2,0.49059200286865234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,16,power_law_1.2,0.502784013748169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,16,power_law_1.2,5.182432174682617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,16,power_law_1.2,0.5017983913421631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,16,power_law_1.2,0.5484416007995605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,16,power_law_1.01,5.495961761474609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,16,power_law_1.2,0.5663680076599121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,16,1,power_law_1.2,1.9027904510498046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,16,1,balanced,0.07489066819349925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,16,1,balanced,0.13583466410636902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,16,power_law_1.2,0.6159488201141358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,16,1,balanced,0.24930665890375772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,16,1,balanced,0.48633599281311035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,16,1,balanced,0.9221920172373453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,16,power_law_1.2,0.6664896011352539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,16,1,balanced,1.3445760409037273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,16,power_law_1.2,0.6979455947875977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,16,1,balanced,1.588314692179362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,16,power_law_1.2,0.8002047538757324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,16,1,balanced,1.5955999692281086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,16,power_law_1.01,7.611103820800781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,16,power_law_1.2,0.9691519737243652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,16,power_law_1.2,6.779090881347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,16,power_law_1.2,1.1789312362670898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,16,power_law_1.2,1.5784704208374023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,16,1,balanced,7.561605453491211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,16,power_law_1.2,1.9521984100341796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,16,1,balanced,0.32629867394765216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,16,1,balanced,0.08217599987983704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,16,1,balanced,0.13166933258374533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,16,1,balanced,0.32653866211573285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,16,1,balanced,0.2415999968846639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,16,1,balanced,7.564048131306966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,16,1,balanced,0.33133333921432495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,16,power_law_1.2,2.719046401977539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,16,1,balanced,0.33268799384435016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,16,1,balanced,0.3343573411305745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,16,1,balanced,0.3371093273162842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,16,power_law_1.2,3.691743850708008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,16,1,balanced,6.460357030232747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,16,1,balanced,2.2797226905822754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,16,1,balanced,2.2899680137634277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,16,power_law_1.2,5.145036697387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,16,1,balanced,2.300656000773112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,16,1,balanced,2.3120800654093423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,16,1,power_law_1.01,0.07434239983558655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,16,1,balanced,2.328762690226237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,16,1,power_law_1.01,0.13157119750976562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,16,1,balanced,2.3554293314615884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,16,1,power_law_1.01,0.24713599681854248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,16,power_law_1.2,6.6086784362792965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,16,1,balanced,6.121994654337565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,16,1,balanced,0.5139040152231852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,16,1,balanced,0.9736586411794027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,16,1,power_law_1.01,0.07336959838867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,16,1,balanced,1.4255733489990234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,16,1,balanced,6.163834889729817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,16,1,power_law_1.01,0.29410560131073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,16,1,power_law_1.01,0.13023359775543214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,16,1,power_law_1.01,0.29184000492095946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,16,1,power_law_1.01,0.23839359283447265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,16,1,power_law_1.01,0.30913920402526857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,16,1,power_law_1.01,0.32626559734344485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,16,1,power_law_1.01,0.33346560001373293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,16,1,balanced,6.261482874552409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,16,1,power_law_1.01,0.35778560638427737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,16,power_law_1.2,9.708799743652344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,16,1,balanced,6.380746841430664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,16,1,power_law_1.2,0.8372287750244141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,16,1,power_law_1.01,0.43053441047668456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,16,1,power_law_1.01,0.611840009689331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,16,1,balanced,11.14560572306315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,16,1,power_law_1.01,1.0334336280822753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,16,1,balanced,11.301157633463541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,16,1,balanced,1.5705653826395671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,16,1,power_law_1.01,0.44720001220703126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,16,1,balanced,1.5753919283548992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,16,1,power_law_1.01,0.6372543811798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,16,power_law_1.2,12.435814666748048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,16,1,power_law_1.01,1.0677056312561035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,16,1,balanced,16.36077372233073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,16,1,power_law_1.2,0.07182719707489013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,16,1,power_law_1.2,0.1306175947189331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,16,1,power_law_1.2,0.23876481056213378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,16,1,power_law_1.2,0.4197375774383545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,16,1,power_law_1.2,0.5828288078308106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,16,1,power_law_1.01,1.2938624382019044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,16,1,power_law_1.2,1.0707584381103517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,16,1,power_law_1.01,1.3135680198669433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,16,1,balanced,17.410847981770832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,16,1,power_law_1.2,1.2531840324401855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,16,1,power_law_1.2,1.2944128036499023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,16,1,balanced,1.8138987223307292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,16,1,power_law_1.2,1.5563776016235351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,16,1,balanced,1.8126932779947917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,16,1,power_law_1.2,1.573305606842041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,16,1,power_law_1.01,1.2406911849975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,16,1,power_law_1.2,1.5149120330810546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,16,1,power_law_1.01,1.321561622619629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,16,1,power_law_1.2,1.3372480392456054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,16,1,power_law_1.2,1.3724096298217774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,16,1,power_law_1.2,1.406380844116211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,16,1,power_law_1.2,1.464691162109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,16,1,power_law_1.2,1.5225600242614745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,16,1,balanced,27.705973307291668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,16,1,power_law_1.2,1.6337215423583984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,16,1,power_law_1.2,0.3023871898651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,16,1,power_law_1.2,0.3039360046386719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,16,1,balanced,0.7833759784698486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,16,1,power_law_1.2,1.727846336364746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,16,1,power_law_1.2,0.3290816068649292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,16,1,balanced,0.791914701461792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,16,1,power_law_1.2,0.33658881187438966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,16,1,balanced,0.8094879786173502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,16,1,power_law_1.2,0.3440191984176636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,16,1,balanced,0.8255733648935953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,16,1,power_law_1.2,1.8699455261230469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,16,1,power_law_1.2,0.38151040077209475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,16,1,balanced,1.522981325785319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,16,1,balanced,1.5582399368286133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,16,1,power_law_1.2,2.149286460876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,16,1,balanced,2.281498591105143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,16,1,power_law_1.2,2.430476760864258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,16,1,balanced,2.3569493293762207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,16,1,power_law_1.2,2.9891199111938476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,16,1,balanced,3.7860854466756186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,16,1,power_law_1.01,0.805356788635254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,16,1,balanced,4.582912127176921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,16,1,power_law_1.2,3.571014404296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,16,1,power_law_1.01,0.8652416229248047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,16,1,power_law_1.2,0.07370240092277527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,16,1,power_law_1.01,1.0039487838745118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,16,1,power_law_1.2,0.13495680093765258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,16,1,power_law_1.01,1.136531162261963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,16,1,power_law_1.2,0.24739840030670165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,16,1,power_law_1.01,1.3971136093139649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,16,1,power_law_1.2,0.38899199962615966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,16,1,power_law_1.2,4.744710540771484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,16,1,balanced,34.31963857014974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,16,1,power_law_1.01,1.5297344207763672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,32,1,balanced,0.0271573339899381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,32,1,balanced,0.029504001140594482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,32,1,balanced,0.03738666574160258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,32,1,balanced,0.037717332442601524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,32,1,balanced,0.038021333515644073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,32,1,balanced,0.03975466638803482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,32,1,balanced,0.04584533472855886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,32,1,balanced,0.044677332043647766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,16,1,power_law_1.01,1.6700544357299805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,32,1,balanced,0.06029866635799408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,32,1,balanced,0.06047999858856201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,32,1,balanced,0.06012799839178721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,16,1,power_law_1.01,1.5719103813171387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,32,1,balanced,0.06025599936644236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,32,1,balanced,0.060080001751581825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,32,1,balanced,0.06670933465162913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,32,1,balanced,0.08038400113582611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,32,1,balanced,0.08075733482837677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,32,1,balanced,0.07438933352629344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,32,1,balanced,0.08079466720422109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,32,1,balanced,0.0839519997437795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,16,1,power_law_1.01,2.2033407211303713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,32,1,balanced,0.10231467088063557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,32,1,balanced,0.11409067114194234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,32,1,balanced,0.1719520092010498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,32,1,balanced,0.20013866821924844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,32,1,balanced,0.2805386583010356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,16,1,power_law_1.2,5.907807922363281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,32,1,balanced,0.36238932609558105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,32,1,balanced,0.5013226668039957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,16,1,power_law_1.01,2.72805118560791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,16,1,power_law_1.2,0.5635263919830322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,16,1,power_law_1.2,1.0140159606933594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,16,1,power_law_1.01,3.7861953735351563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,16,1,power_law_1.2,1.2546751976013184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,16,1,balanced,1.70907195409139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,32,1,balanced,0.6640586853027344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,32,1,balanced,0.025050667424996693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,16,1,power_law_1.2,1.2979392051696776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,32,1,balanced,0.029504001140594482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,32,1,balanced,0.03142400085926056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,32,1,balanced,0.029578665892283123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,32,1,balanced,0.02942399928967158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,16,1,power_law_1.2,8.315769958496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,32,1,balanced,0.03180799881617228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,32,1,balanced,0.03562133262554804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,32,1,balanced,0.0377813329299291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,32,1,balanced,0.044112001856168113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,32,1,balanced,0.045791998505592346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,32,1,balanced,0.03793066740036011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,32,1,balanced,0.04041066765785217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,32,1,balanced,0.04163199911514918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,32,1,balanced,0.042090664307276406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,16,1,power_law_1.01,4.864038467407227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,32,1,balanced,0.049925332268079124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,16,1,power_law_1.01,6.289388656616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,32,1,balanced,0.052282666166623436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,32,1,balanced,0.050026665131251015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,32,1,balanced,0.06440000236034393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,32,1,balanced,0.06949866811434428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,32,1,balanced,0.08686932921409607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,32,1,balanced,0.10109333197275798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,32,1,balanced,0.14524267117182413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,32,1,balanced,0.17293334007263184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,32,1,balanced,0.24167466163635254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,32,1,balanced,0.3057066599527995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,32,1,balanced,0.4360640048980713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,16,1,power_law_1.2,6.330361557006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,32,1,balanced,0.5626346667607626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,32,1,power_law_1.01,0.024665600061416625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,32,1,power_law_1.01,0.03111039996147156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,32,1,power_law_1.01,0.03288959860801697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,32,1,power_law_1.01,0.036396801471710205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,32,1,power_law_1.01,0.036211198568344115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,16,1,power_law_1.01,6.522438049316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,32,1,power_law_1.01,0.037894400954246524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,32,1,balanced,0.025231999655564625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,32,1,power_law_1.01,0.044588801264762876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,32,1,balanced,0.025424001117547352
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,32,1,balanced,0.02548266698916753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,32,1,balanced,0.025263999899228413
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,32,1,balanced,0.025493333737055462
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,32,1,balanced,0.026885333160559338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,32,1,balanced,0.029189333319664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,32,1,balanced,0.029530666768550873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,32,1,balanced,0.029631999631722767
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,32,1,balanced,0.0303413321574529
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,32,1,balanced,0.032298666735490165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,16,1,power_law_1.2,6.458553314208984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,32,1,balanced,0.034714666505654655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,32,1,balanced,0.033770665526390076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,32,1,balanced,0.03737066686153412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,32,1,balanced,0.04363733530044556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,32,1,balanced,0.041802664597829185
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,32,1,balanced,0.0415786678592364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,32,1,balanced,0.05686933298905691
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,32,1,balanced,0.05524266759554545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,32,1,balanced,0.07229866584142049
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,32,1,balanced,0.08609599868456523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,32,1,balanced,0.11121599872907002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,32,1,balanced,0.1381119986375173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,32,1,power_law_1.01,0.04700160026550293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,32,1,balanced,0.19403199354807535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,32,1,balanced,0.25044800837834674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,32,1,power_law_1.01,0.06223359704017639
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,32,1,balanced,0.3599413235982259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,32,1,power_law_1.01,0.06380800008773804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,16,1,power_law_1.2,5.435251235961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,32,1,balanced,0.46530667940775555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,32,1,power_law_1.01,0.0604095995426178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,32,1,power_law_1.01,0.02305919975042343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,32,1,power_law_1.01,0.026284798979759216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,32,1,power_law_1.01,0.06596480011940002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,32,1,power_law_1.01,0.026950401067733765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,32,1,power_law_1.01,0.06712960004806519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,32,1,power_law_1.01,0.028697600960731505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,32,1,power_law_1.01,0.070796799659729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,32,1,power_law_1.01,0.028774398565292358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,32,1,power_law_1.01,0.029894399642944335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,32,1,power_law_1.01,0.034790399670600894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,32,1,power_law_1.01,0.07861760258674622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,32,1,power_law_1.01,0.036499199271202085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,32,1,power_law_1.01,0.08232319951057435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,32,1,power_law_1.01,0.043968001008033754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,16,1,balanced,1.4463040033976238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,32,1,power_law_1.01,0.07454079985618592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,32,1,power_law_1.01,0.04469760060310364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,32,1,power_law_1.01,0.03771519958972931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,16,1,balanced,1.4589120546976726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,32,1,power_law_1.01,0.08244479894638061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,32,1,power_law_1.01,0.040505599975585935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,32,1,power_law_1.01,0.09098240137100219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,32,1,power_law_1.01,0.04173440039157868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,16,1,balanced,1.468272050221761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,16,1,power_law_1.2,2.0873344421386717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,16,1,balanced,1.4784693717956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,16,1,balanced,1.4905227025349934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,16,1,power_law_1.2,2.1467456817626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,16,1,balanced,1.5162933667500813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,32,1,power_law_1.01,0.10636160373687745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,32,1,power_law_1.01,0.12103040218353271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,16,1,power_law_1.2,2.2106752395629883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,32,1,power_law_1.01,0.044582399725914004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,32,1,power_law_1.01,0.05122560262680054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,32,1,power_law_1.01,0.057606399059295654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,16,1,power_law_1.2,2.3171072006225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,32,1,power_law_1.01,0.05242239832878113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,16,1,power_law_1.2,2.418771171569824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,32,1,power_law_1.01,0.06439679861068726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,32,1,power_law_1.01,0.1759168028831482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,16,1,power_law_1.2,2.5540224075317384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,16,1,power_law_1.01,1.4846400260925292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,32,1,power_law_1.01,0.07288960218429566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,16,1,power_law_1.2,10.684127807617188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,32,1,power_law_1.01,0.09187840223312378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,16,1,power_law_1.2,6.462841796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,32,1,power_law_1.01,0.10426239967346192
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,32,1,power_law_1.01,0.023014399409294128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,32,1,power_law_1.01,0.023449599742889404
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,32,1,power_law_1.01,0.024255999922752382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,16,1,power_law_1.2,0.8506367683410645
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,32,1,power_law_1.01,0.02465279996395111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,32,1,power_law_1.01,0.02476159930229187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,16,1,power_law_1.2,0.9067647933959961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,32,1,power_law_1.01,0.026259198784828186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,32,1,power_law_1.01,0.027647998929023743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,16,1,power_law_1.2,1.0283007621765137
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,32,1,power_law_1.01,0.028774398565292358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,16,1,power_law_1.2,6.916255950927734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,32,1,power_law_1.01,0.029049599170684816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,32,1,power_law_1.01,0.02956799864768982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,16,1,power_law_1.2,1.1619199752807616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,32,1,power_law_1.01,0.032358399033546446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,32,1,power_law_1.01,0.03290880024433136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,16,1,power_law_1.2,1.422208023071289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,32,1,power_law_1.01,0.03333120048046112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,32,1,power_law_1.01,0.036595198512077334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,16,1,balanced,1.6477173169453938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,32,1,power_law_1.01,0.0377023994922638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,16,1,power_law_1.2,1.6934591293334962
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,32,1,power_law_1.01,0.03984639942646027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,16,1,balanced,1.6879839897155762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,32,1,power_law_1.01,0.04185599982738495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,16,1,balanced,1.7701706886291504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,16,1,power_law_1.2,7.748358154296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,16,1,power_law_1.2,2.219647979736328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,32,1,power_law_1.01,0.14687360525131227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,16,1,balanced,1.8562240600585938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,32,1,power_law_1.01,0.051596802473068235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,32,1,power_law_1.01,0.05904639959335327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,16,1,power_law_1.2,2.748588752746582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,16,1,balanced,3.2189547220865884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,32,1,power_law_1.01,0.07265920042991639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,32,1,power_law_1.01,0.20523519515991212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,16,1,power_law_1.01,1.295468807220459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,16,1,balanced,3.3917067845662436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,16,1,power_law_1.2,3.8281280517578127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,16,1,power_law_1.01,1.3378751754760743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,16,1,power_law_1.2,8.716089630126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,16,1,power_law_1.01,1.351039981842041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,16,1,balanced,4.9570614496866865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,16,1,power_law_1.01,1.4166144371032714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,16,1,power_law_1.01,1.471008014678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,16,1,balanced,5.317610740661621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,32,1,power_law_1.01,0.28513278961181643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,16,1,power_law_1.01,1.5641599655151368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,32,1,power_law_1.01,0.17889920473098755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,32,1,power_law_1.01,0.08476160168647766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,16,1,power_law_1.2,4.901055908203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,32,1,power_law_1.2,0.02502399981021881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,32,1,power_law_1.2,0.02930560111999512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,32,1,power_law_1.2,0.031430399417877196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,32,1,power_law_1.2,0.035769599676132205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,32,1,power_law_1.2,0.035769599676132205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,16,1,balanced,8.654794692993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,32,1,power_law_1.2,0.03814400136470795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,32,1,power_law_1.2,0.044172799587249754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,32,1,power_law_1.2,0.04640640020370483
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,32,1,power_law_1.01,0.11112320423126221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,32,1,power_law_1.2,0.061318397521972656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,32,1,power_law_1.2,0.06331520080566407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,16,1,power_law_1.2,10.53512954711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,32,1,power_law_1.2,0.061638402938842776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,32,1,power_law_1.2,0.06421120166778564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,32,1,power_law_1.2,0.06533759832382202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,32,1,power_law_1.2,0.07030400037765502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,16,1,balanced,10.4115842183431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,32,1,power_law_1.2,0.07618560194969178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,32,1,power_law_1.2,0.08147199749946595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,32,1,power_law_1.2,0.07453439831733703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,32,1,power_law_1.01,0.13797760009765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,16,1,power_law_1.2,12.424658966064452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,32,1,power_law_1.2,0.0822655975818634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,32,1,power_law_1.2,0.09099519848823548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,32,1,power_law_1.2,0.023904000222682954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,32,1,power_law_1.2,0.027616000175476073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,32,1,power_law_1.2,0.02911359965801239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,32,1,power_law_1.2,0.029555198550224305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,32,1,power_law_1.2,0.029548799991607665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,32,1,power_law_1.2,0.030931198596954347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,32,1,power_law_1.2,0.035692799091339114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,32,1,power_law_1.01,0.24700160026550294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,32,1,power_law_1.2,0.03697920143604279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,32,1,power_law_1.2,0.04501760005950928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,32,1,power_law_1.2,0.04549759924411774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,32,1,power_law_1.2,0.039001598954200745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,16,1,power_law_1.01,1.6746431350708009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,16,1,power_law_1.01,1.7852159500122071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,32,1,power_law_1.01,0.3692863941192627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,16,1,power_law_1.2,16.227052307128908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,16,1,power_law_1.01,2.0714431762695313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,32,1,power_law_1.01,0.19530240297317505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,32,1,power_law_1.2,0.10748800039291381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,16,1,power_law_1.01,2.356902313232422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,16,1,power_law_1.01,2.92873592376709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,16,1,power_law_1.01,3.4920894622802736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,16,1,power_law_1.01,4.65494384765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,16,1,power_law_1.01,5.6266624450683596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,16,1,power_law_1.2,20.003648376464845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,32,1,power_law_1.2,0.12066559791564942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,16,1,power_law_1.01,5.821241760253907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,32,1,power_law_1.2,0.041075199842453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,32,1,power_law_1.01,0.31140480041503904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,32,1,power_law_1.2,0.04298880100250244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,32,1,power_law_1.2,0.043961599469184875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,32,1,power_law_1.01,0.25445759296417236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,16,1,power_law_1.01,8.201171112060546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,32,1,power_law_1.2,0.05002239942550659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,32,1,power_law_1.2,0.05333120226860046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,32,1,power_law_1.2,0.054124802350997925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,32,1,power_law_1.01,0.51080322265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,16,1,power_law_1.01,10.5849853515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,32,1,power_law_1.2,0.1785215973854065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,16,1,power_law_1.2,27.866183471679687
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,32,1,power_law_1.2,0.024607999622821806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,32,1,power_law_1.2,0.025139200687408447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,32,1,power_law_1.2,0.025248000025749208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,32,1,power_law_1.2,0.025446400046348572
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,32,1,power_law_1.2,0.025779199600219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,32,1,power_law_1.2,0.026745599508285523
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,32,1,power_law_1.2,0.028262400627136232
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,32,1,power_law_1.2,0.029951998591423036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,32,1,power_law_1.2,0.029414400458335876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,32,1,power_law_1.2,0.02995840013027191
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,32,1,power_law_1.2,0.031167998909950256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,32,1,power_law_1.2,0.03237760066986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,32,1,power_law_1.2,0.033939200639724734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,32,1,power_law_1.2,0.036025598645210266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,16,1,power_law_1.01,2.031462478637695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,32,1,power_law_1.2,0.03715200126171112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,16,1,power_law_1.01,2.0806400299072267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,32,1,power_law_1.2,0.038867199420928956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,32,1,power_law_1.2,0.04193280041217804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,16,1,power_law_1.01,2.1008575439453123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,16,1,power_law_1.01,2.177926445007324
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,32,1,power_law_1.2,0.051846402883529666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,16,1,power_law_1.01,2.2908607482910157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,32,1,power_law_1.2,0.2084415912628174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,16,1,power_law_1.01,2.4351871490478514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,32,1,power_law_1.2,0.06377599835395813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,32,1,power_law_1.2,0.07310720086097718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,32,1,power_law_1.2,0.060524797439575194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,32,1,power_law_1.01,0.4451903820037842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,16,1,power_law_1.2,35.637228393554686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,32,1,balanced,0.03151999910672506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,32,1,balanced,0.03851199895143509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,32,1,balanced,0.05309866865475973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,32,1,balanced,0.053029333551724754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,32,1,balanced,0.052245333790779114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,32,1,balanced,0.05339199801286062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,32,1,balanced,0.06474133332570393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,32,1,balanced,0.06576000154018402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,32,1,balanced,0.08704533179601033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,32,1,balanced,0.08656000097592671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,32,1,balanced,0.08689066767692566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,32,1,balanced,0.08753599723180135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,32,1,balanced,0.09045333663622539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,32,1,balanced,0.09269866347312927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,32,1,balanced,0.1030453344186147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,32,1,balanced,0.10730133454004924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,32,1,balanced,0.09886399904886882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,32,1,balanced,0.11342933773994446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,32,1,balanced,0.1172106663386027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,32,1,balanced,0.1420960028966268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,32,1,balanced,0.16265599926312765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,32,1,balanced,0.2463093400001526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,32,1,balanced,0.29020265738169354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,32,1,balanced,0.41411733627319336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,32,1,balanced,0.545194665590922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,32,1,balanced,0.7902613480885824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,32,1,power_law_1.01,0.35976319313049315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,32,1,balanced,1.0071679751078289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,32,1,balanced,0.029264000554879505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,32,1,balanced,0.032287999987602234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,32,1,balanced,0.03532266616821289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,32,1,balanced,0.03578133384386698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,32,1,balanced,0.03562666724125544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,32,1,balanced,0.03740799923737844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,32,1,balanced,0.04590400060017904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,32,1,balanced,0.0455626646677653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,32,1,balanced,0.05806399881839752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,32,1,balanced,0.05901333192984263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,32,1,balanced,0.05169600248336792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,32,1,power_law_1.2,0.09098240137100219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,32,1,balanced,0.054144000013669334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,32,1,balanced,0.05786666770776113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,32,1,power_law_1.2,0.0723136007785797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,32,1,balanced,0.06405866642793019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,32,1,balanced,0.07910933097203572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,32,1,balanced,0.08025066554546356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,32,1,balanced,0.07419200241565704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,32,1,balanced,0.08707732955614726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,32,1,balanced,0.09268266956011455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,32,1,balanced,0.11912000179290771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,32,1,balanced,0.13434132933616638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,32,1,balanced,0.19362133741378784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,32,1,balanced,0.23452266057332358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,32,1,balanced,0.33234665791193646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,32,1,balanced,0.4352586666742961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,32,1,balanced,0.6217120091120402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,32,1,balanced,0.8038826783498129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,32,1,balanced,0.041877334316571556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,32,1,balanced,0.047925333182017006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,32,1,balanced,0.055861334005991616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,32,1,balanced,0.055904000997543335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,32,1,balanced,0.057904000083605446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,32,1,balanced,0.058245331048965454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,32,1,balanced,0.05919999877611796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,32,1,balanced,0.060138667623202004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,32,1,balanced,0.09275733431180318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,32,1,balanced,0.09095999598503113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,32,1,balanced,0.09150933225949605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,32,1,balanced,0.09689066807428996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,32,1,balanced,0.09943466385205586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,32,1,balanced,0.10152533650398254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,32,1,balanced,0.1178559958934784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,32,1,balanced,0.1188320020834605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,32,1,balanced,0.11488533020019531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,32,1,balanced,0.13362133502960205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,32,1,balanced,0.14888532956441244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,32,1,balanced,0.23161067565282187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,32,1,balanced,0.2659413417180379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,32,1,balanced,0.37696532408396405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,32,1,balanced,0.49347734451293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,32,1,balanced,0.6926133632659912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,32,1,balanced,0.9059253533681234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,32,1,balanced,1.2943519751230876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,32,1,balanced,1.6952959696451824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,32,1,power_law_1.2,0.10508160591125489
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,32,1,power_law_1.2,0.0843455970287323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,32,1,power_law_1.2,0.2877887964248657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,32,1,balanced,0.025546667476495106
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,32,1,balanced,0.027514666318893433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,32,1,balanced,0.029264000554879505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,32,1,balanced,0.029253333806991577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,32,1,balanced,0.029285334050655365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,32,1,balanced,0.02926933268706004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,32,1,balanced,0.03133333226044973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,32,1,balanced,0.0332640012105306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,32,1,balanced,0.03339199970165888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,32,1,balanced,0.03505599995454153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,32,1,balanced,0.035775999228159584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,32,1,balanced,0.038191998998324074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,32,1,balanced,0.0396373321612676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,32,1,balanced,0.04181866844495138
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,32,1,balanced,0.04837333162625631
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,32,1,balanced,0.0491839994986852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,32,1,balanced,0.048725331823031105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,32,1,balanced,0.07470400134722392
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,32,1,balanced,0.0746506651242574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,32,1,balanced,0.09341866771380107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,32,1,balanced,0.1090133289496104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,32,1,balanced,0.14833066860834757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,32,1,balanced,0.19001599152882895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,32,1,balanced,0.2712693413098653
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,32,1,balanced,0.3452106714248657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,32,1,balanced,0.5022559960683187
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,32,1,balanced,0.6526399850845337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,32,1,power_law_1.01,0.0335999995470047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,32,1,power_law_1.01,0.039724799990653994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,32,1,power_law_1.01,0.04380159974098206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,32,1,power_law_1.01,0.05013120174407959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,32,1,power_law_1.01,0.05112959742546082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,32,1,power_law_1.01,0.05319039821624756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,32,1,power_law_1.01,0.0639360010623932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,32,1,power_law_1.01,0.06670079827308655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,32,1,power_law_1.01,0.08610560297966004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,32,1,power_law_1.01,0.08687360286712646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,32,1,power_law_1.01,0.6781888008117676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,32,1,power_law_1.01,0.08851839900016785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,32,1,power_law_1.01,0.027910399436950683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,32,1,power_law_1.2,0.14895360469818114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,32,1,power_law_1.01,0.09241600036621093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,16,1,power_law_1.01,6.149343872070313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,32,1,power_law_1.01,0.03318400084972382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,32,1,power_law_1.01,0.09409919977188111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,32,1,power_law_1.01,0.03373439908027649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,32,1,power_law_1.01,0.03653120100498199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,32,1,power_law_1.01,0.09722239971160888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,32,1,power_law_1.01,0.034841600060462954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,32,1,power_law_1.01,0.10308480262756348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,32,1,power_law_1.01,0.041382399201393125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,32,1,power_law_1.01,0.04705919921398163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,32,1,power_law_1.01,0.05287039875984192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,32,1,power_law_1.01,0.10610560178756714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,32,1,power_law_1.01,0.06124160289764404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,32,1,power_law_1.01,0.06219519972801209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,32,1,power_law_1.01,0.10280319452285766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,32,1,power_law_1.01,0.05910400152206421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,32,1,power_law_1.01,0.11641600131988525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,32,1,power_law_1.01,0.06305919885635376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,16,1,power_law_1.01,6.616397094726563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,32,1,power_law_1.01,0.06593919992446899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,32,1,power_law_1.01,0.0716480016708374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,32,1,power_law_1.01,0.12736639976501465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,32,1,power_law_1.01,0.07685760259628296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,32,1,power_law_1.01,0.08179200291633607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,32,1,power_law_1.01,0.07837439775466919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,32,1,power_law_1.01,0.08860160112380981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,32,1,power_law_1.01,0.5808063983917237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,32,1,power_law_1.01,0.04156160056591034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,32,1,power_law_1.01,0.05077120065689087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,16,1,power_law_1.01,7.447795104980469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,32,1,power_law_1.01,0.05359359979629517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,32,1,power_law_1.01,0.05449600219726562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,32,1,power_law_1.01,0.057004797458648684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,32,1,power_law_1.01,0.0582144021987915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,32,1,power_law_1.01,0.09962239861488342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,32,1,power_law_1.01,0.0572160005569458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,32,1,power_law_1.01,0.06262400150299072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,32,1,power_law_1.01,0.09200639724731445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,32,1,power_law_1.01,0.1208448052406311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,32,1,power_law_1.01,0.09168639779090881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,32,1,power_law_1.2,0.11041280031204223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,32,1,power_law_1.01,0.0942463994026184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,32,1,power_law_1.01,0.09744639992713929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,32,1,power_law_1.01,0.09750400185585022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,32,1,power_law_1.01,0.10704640150070191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,16,1,power_law_1.01,8.344863891601562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,32,1,power_law_1.01,0.11423360109329224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,32,1,power_law_1.01,0.1390015959739685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,32,1,power_law_1.01,0.12028800249099732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,32,1,power_law_1.01,0.1506495952606201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,32,1,power_law_1.01,0.11927039623260498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,32,1,power_law_1.01,0.20112640857696534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,32,1,power_law_1.2,0.1792255997657776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,32,1,power_law_1.01,0.1364799976348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,16,1,power_law_1.01,10.19475860595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,32,1,power_law_1.01,0.1856063961982727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,32,1,power_law_1.01,0.1525760054588318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,32,1,power_law_1.01,0.23629438877105713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,32,1,power_law_1.01,0.2589375972747803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,32,1,power_law_1.01,0.297760009765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,16,1,power_law_1.01,12.019014739990235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,32,1,power_law_1.01,0.4668735980987549
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,32,1,power_law_1.01,0.2691391944885254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,32,1,power_law_1.01,0.37798399925231935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,32,1,power_law_1.2,0.37088000774383545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,32,1,power_law_1.01,0.024825599789619446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,32,1,power_law_1.01,0.025446400046348572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,32,1,power_law_1.01,0.23919999599456787
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,32,1,power_law_1.01,0.02624639868736267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,32,1,power_law_1.01,0.027167999744415285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,32,1,power_law_1.01,0.027404800057411194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,32,1,power_law_1.2,0.1368191957473755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,32,1,power_law_1.01,0.48487038612365724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,16,1,power_law_1.01,15.781382751464843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,32,1,power_law_1.01,0.028966400027275085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,32,1,power_law_1.01,0.03031040132045746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,32,1,power_law_1.01,0.0335999995470047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,32,1,power_law_1.01,0.032332798838615416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,32,1,power_law_1.01,0.03344640135765076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,32,1,power_law_1.01,0.036774399876594546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,32,1,power_law_1.01,0.03615359961986542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,32,1,power_law_1.01,0.03824639916419983
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,32,1,power_law_1.01,0.04437119960784912
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,32,1,power_law_1.01,0.04255360066890716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,32,1,power_law_1.01,0.42290558815002444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,32,1,power_law_1.01,0.04460160136222839
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,32,1,power_law_1.01,0.05242879986763001
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,32,1,power_law_1.01,0.0634112000465393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,32,1,power_law_1.01,0.07781760096549988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,16,1,power_law_1.01,19.552198791503905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,32,1,power_law_1.01,0.0942463994026184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,32,1,power_law_1.01,0.33596799373626707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,32,1,power_law_1.01,0.11129599809646606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,32,1,power_law_1.01,0.6823296070098877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,32,1,power_law_1.2,0.24598400592803954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,32,1,power_law_1.01,0.14954240322113038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,16,1,power_law_1.01,27.432928466796874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,32,1,power_law_1.01,0.19137920141220094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,32,1,power_law_1.01,0.5509632110595704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,32,1,power_law_1.2,0.1942463994026184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,16,1,power_law_1.01,35.324459838867185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,32,1,power_law_1.2,0.03529599905014038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,32,1,power_law_1.2,0.03973119854927063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,32,1,power_law_1.2,0.04312320053577423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,32,1,power_law_1.2,0.05004159808158874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,32,1,power_law_1.2,0.05134720206260681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,32,1,power_law_1.2,0.05246719717979431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,32,1,power_law_1.2,0.06396160125732422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,32,1,power_law_1.2,0.06664320230484008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,32,1,power_law_1.2,0.08604159951210022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,32,1,power_law_1.2,0.08679680228233337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,32,1,power_law_1.01,0.43653120994567873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,32,1,power_law_1.2,0.08753920197486878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,32,1,power_law_1.2,0.09268479943275451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,32,1,power_law_1.2,0.0937279999256134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,32,1,power_law_1.2,0.09619839787483216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,32,1,power_law_1.2,0.10240639448165893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,32,1,power_law_1.2,0.10557440519332886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,32,1,power_law_1.2,0.10324480533599853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,32,1,power_law_1.2,0.11593600511550903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,32,1,power_law_1.2,0.31334400177001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,32,1,power_law_1.01,0.27370879650115965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,32,1,power_law_1.2,0.12817280292510985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,32,1,power_law_1.2,0.14869120121002197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,32,1,power_law_1.01,0.8753727912902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,32,1,power_law_1.2,0.17855360507965087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,32,1,power_law_1.2,0.2562880039215088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,32,1,power_law_1.2,0.25311999320983886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,32,1,power_law_1.2,0.5143551826477051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,32,1,power_law_1.01,0.794975996017456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,32,1,power_law_1.01,0.6364799976348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,32,1,power_law_1.01,1.2750399589538575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,32,1,power_law_1.01,0.3519871950149536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,32,1,power_law_1.2,0.6774400234222412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,32,1,power_law_1.2,0.02810879945755005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,32,1,power_law_1.2,0.2929088115692139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,32,1,power_law_1.2,0.032915198802948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,32,1,power_law_1.2,0.03454079926013946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,32,1,power_law_1.2,0.035872000455856326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,32,1,power_law_1.2,0.035641598701477054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,32,1,power_law_1.2,0.038335999846458434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,32,1,power_law_1.2,0.045184001326560974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,32,1,power_law_1.2,0.04745599925518036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,32,1,power_law_1.2,0.05983359813690185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,32,1,power_law_1.2,0.06165760159492493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,32,1,power_law_1.2,0.05509120225906372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,32,1,power_law_1.2,0.06273919939994813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,32,1,power_law_1.2,0.0651199996471405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,32,1,power_law_1.2,0.07128959894180298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,32,1,power_law_1.2,0.07694720029830933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,32,1,power_law_1.2,0.08165119886398316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,32,1,power_law_1.2,0.07716479897499084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,32,1,power_law_1.2,0.08859519958496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,32,1,power_law_1.2,0.09844480156898498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,32,1,power_law_1.01,0.5029695987701416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,32,1,power_law_1.2,0.11973119974136352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,32,1,power_law_1.2,0.42525439262390136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,32,1,power_law_1.2,0.14410239458084106
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,32,1,power_law_1.2,0.3613055944442749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,32,1,power_law_1.2,0.20069758892059325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,32,1,power_law_1.01,1.6733760833740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,32,1,power_law_1.2,0.0419840008020401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,32,1,power_law_1.2,0.048851200938224794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,32,1,power_law_1.2,0.054022401571273804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,32,1,power_law_1.2,0.055692797899246214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,32,1,power_law_1.2,0.056569600105285646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,32,1,power_law_1.2,0.056883198022842404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,32,1,power_law_1.2,0.056543999910354616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,32,1,power_law_1.2,0.05947520136833191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,32,1,power_law_1.2,0.08936319947242737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,32,1,power_law_1.2,0.08823040127754211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,32,1,power_law_1.2,0.08974720239639282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,32,1,power_law_1.2,0.09282559752464295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,32,1,power_law_1.2,0.096806401014328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,32,1,power_law_1.2,0.1072383999824524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,32,1,power_law_1.2,0.1139456033706665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,32,1,power_law_1.2,0.11990400552749633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,32,1,power_law_1.2,0.12002559900283813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,32,1,power_law_1.2,0.13638399839401244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,32,1,power_law_1.2,0.44223999977111816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,32,1,power_law_1.2,0.5475264072418213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,32,1,power_law_1.2,0.2442944049835205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,32,1,power_law_1.2,0.1518720030784607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,32,1,power_law_1.01,1.0440768241882323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,32,1,power_law_1.2,0.23226239681243896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,32,1,power_law_1.2,0.024358400702476503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,32,1,power_law_1.2,0.025727999210357667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,32,1,power_law_1.2,0.02600319981575012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,32,1,power_law_1.2,0.027302399277687073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,32,1,power_law_1.2,0.027238398790359497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,32,1,power_law_1.2,0.028569599986076354
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,32,1,power_law_1.2,0.03038719892501831
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,32,1,power_law_1.2,0.033932799100875856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,32,1,power_law_1.2,0.032543998956680295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,32,1,power_law_1.2,0.03325439989566803
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,32,1,power_law_1.2,0.03706879913806915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,32,1,power_law_1.2,0.03623040020465851
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,32,1,power_law_1.2,0.038771200180053714
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,32,1,power_law_1.2,0.044512000679969785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,32,1,power_law_1.2,0.2685312032699585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,32,1,power_law_1.2,0.04264959990978241
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,32,1,power_law_1.2,0.045612800121307376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,32,1,power_law_1.2,0.053337597846984865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,32,1,power_law_1.2,0.06496639847755432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,32,1,power_law_1.2,0.3830719947814941
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,32,1,power_law_1.2,0.07915520071983337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,32,1,power_law_1.2,0.09312639832496643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,32,1,power_law_1.01,0.8239808082580566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,32,balanced,0.051962668697039284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,32,balanced,0.051967998345692955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,32,balanced,0.052245333790779114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,32,balanced,0.06611200173695882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,32,balanced,0.09736532966295879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,32,balanced,0.17890133460362753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,32,balanced,0.15738667050997415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,32,balanced,0.16007999579111734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,32,balanced,0.16929600636164346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,32,balanced,0.1720906694730123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,32,balanced,0.23492799202601114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,32,balanced,0.24101332823435465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,32,balanced,0.24744532505671182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,32,balanced,0.2571733395258586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,32,balanced,0.26661332448323566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,32,balanced,0.2751786708831787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,32,balanced,0.29312000672022503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,32,balanced,0.3521493275960286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,32,balanced,0.38115731875101727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,32,1,power_law_1.2,0.8015104293823242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,32,balanced,0.4419519901275635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,32,balanced,0.5026293198267618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,32,balanced,0.6255253156026205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,32,balanced,0.7489706675211588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,32,balanced,1.0700426896413167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,32,balanced,1.3154773712158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,32,balanced,1.8808746337890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,32,balanced,2.451770623524984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,32,balanced,0.040037333965301514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,32,1,power_law_1.2,0.3417279958724976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,32,balanced,0.04186133543650309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,32,balanced,0.04188266893227895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,32,balanced,0.05397866666316986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,32,balanced,0.07072533170382182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,32,balanced,0.10726933677991231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,32,balanced,0.1307253340880076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,32,balanced,0.1323199967543284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,32,balanced,0.13593600193659464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,32,balanced,0.13770133256912231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,32,balanced,0.19124799966812134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,32,balanced,0.2039946715037028
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,32,1,power_law_1.2,0.10956799983978271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,32,balanced,0.20963732401529947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,32,balanced,0.22129066785176596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,32,balanced,0.2312320073445638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,32,balanced,0.24254933993021646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,32,balanced,0.26523733139038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,32,balanced,0.30961066484451294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,32,1,power_law_1.2,0.48287358283996584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,32,balanced,0.3429866631825765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,32,balanced,0.4066186745961507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,32,balanced,0.47650667031606037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,32,1,power_law_1.01,0.6573247909545898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,32,balanced,0.6122879981994629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,32,balanced,0.06227200229962667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,32,balanced,0.06389866769313812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,32,balanced,0.06238399942715963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,32,balanced,0.7460160255432129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,32,balanced,0.07638399799664815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,32,balanced,0.08878399928410848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,32,balanced,0.12082667152086894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,32,balanced,1.060869296391805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,32,balanced,1.3311306635538738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,32,balanced,1.908389409383138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,32,balanced,0.1237333317597707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,32,balanced,0.12454932928085327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,32,balanced,0.23486934105555216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,32,balanced,0.23905066649119058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,32,balanced,0.24665600061416626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,32,balanced,0.2524426579475403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,32,balanced,0.2618933320045471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,32,balanced,0.2752479910850525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,32,balanced,0.28813334306081134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,32,balanced,0.3021226723988851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,32,balanced,0.32892266909281415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,32,1,power_law_1.2,0.15176960229873657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,32,balanced,0.3775893449783325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,32,balanced,0.4294346570968628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,32,balanced,0.5288586616516113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,32,balanced,2.4805119832356772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,32,balanced,0.6299360195795695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,32,balanced,0.9066932996114095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,32,balanced,1.10753067334493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,32,balanced,1.5798932711283367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,32,balanced,2.0573226610819497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,32,balanced,3.0048958460489907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,32,balanced,3.95034122467041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,32,power_law_1.01,0.1527168035507202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,32,power_law_1.01,0.15438719987869262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,32,power_law_1.01,0.15318399667739868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,32,power_law_1.01,0.15560319423675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,32,power_law_1.01,0.1275007963180542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,32,balanced,0.030432000756263733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,32,balanced,0.03136000037193298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,32,power_law_1.01,0.12008320093154908
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,32,balanced,0.03138133386770884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,32,balanced,0.03172266731659571
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,32,balanced,0.04205333193143209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,32,power_law_1.01,0.15418239831924438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,32,power_law_1.01,0.14843519926071166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,32,power_law_1.01,0.15281920433044432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,32,power_law_1.01,0.15544960498809815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,32,power_law_1.01,0.22268800735473632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,32,power_law_1.01,0.22933120727539064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,32,power_law_1.01,0.23921279907226561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,32,power_law_1.01,0.24374399185180665
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,32,balanced,0.06353599826494853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,32,balanced,0.06648533542950948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,32,balanced,0.06863466898600261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,32,power_law_1.01,0.25849599838256837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,32,balanced,0.09934399525324504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,32,balanced,0.10044800241788228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,32,balanced,0.08871466914812724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,32,power_law_1.01,0.26917119026184083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,32,balanced,0.22025066614151
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,32,balanced,0.22152000665664673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,32,power_law_1.01,0.28723199367523194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,32,power_law_1.01,0.35350399017333983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,32,1,power_law_1.2,0.4424704074859619
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,32,1,power_law_1.2,0.466918420791626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,32,power_law_1.01,0.4002816200256348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,32,power_law_1.01,0.0883520007133484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,32,power_law_1.01,0.47649922370910647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,32,power_law_1.01,0.09175040125846863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,32,power_law_1.01,0.5626368045806884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,32,power_law_1.01,0.0924351990222931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,32,power_law_1.01,0.09333760142326356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,32,power_law_1.01,0.7367936134338379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,32,balanced,0.18449066082636514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,32,balanced,0.11743999520937602
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,32,balanced,0.08886399865150452
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,32,balanced,0.09701333443323772
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,32,balanced,0.10493866602579753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,32,balanced,0.11379733681678772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,32,1,power_law_1.2,0.6806848049163818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,32,balanced,0.13214932878812155
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,32,balanced,0.14843199650446573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,32,balanced,0.19056532780329385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,32,balanced,0.2344533403714498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,32,balanced,0.3295146624247233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,32,balanced,0.41380266348520917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,32,balanced,0.5908213456471761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,32,power_law_1.01,0.09221119880676269
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,32,balanced,0.7712106704711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,32,power_law_1.01,0.9056703567504882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,32,power_law_1.01,0.09747840166091919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,32,power_law_1.01,0.10037120580673217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,32,power_law_1.01,0.12012159824371338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,32,power_law_1.01,1.3054207801818847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,32,power_law_1.01,0.10401920080184937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,32,power_law_1.01,0.12186880111694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,32,power_law_1.01,0.10517120361328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,32,power_law_1.01,0.11535359621047973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,32,power_law_1.01,0.10553599596023559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,32,power_law_1.01,1.6150720596313477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,32,power_law_1.01,0.10238720178604126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,32,1,power_law_1.2,0.190937602519989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,32,power_law_1.01,0.11338880062103271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,32,power_law_1.01,0.11841919422149658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,32,power_law_1.01,2.4276735305786135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,32,power_law_1.01,0.1181439995765686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,32,power_law_1.01,0.2141887903213501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,32,power_law_1.01,0.23049600124359132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,32,power_law_1.01,0.12617599964141846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,32,power_law_1.01,0.23505918979644774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,32,power_law_1.01,0.1789888024330139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,32,power_law_1.01,3.1023487091064452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,32,power_law_1.01,0.24376959800720216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,32,power_law_1.01,0.19527039527893067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,32,1,power_law_1.2,0.5759552001953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,32,power_law_1.01,0.2085952043533325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,32,power_law_1.2,0.1522304058074951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,32,power_law_1.01,0.21853439807891845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,32,power_law_1.2,0.1542847990989685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,32,power_law_1.01,0.22677760124206542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,32,power_law_1.2,0.1533184051513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,32,power_law_1.01,0.23415679931640626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,32,power_law_1.2,0.1539072036743164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,32,power_law_1.01,0.2651968002319336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,32,power_law_1.2,0.15455360412597657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,32,power_law_1.01,0.30969600677490233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,32,power_law_1.2,0.1437183976173401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,32,power_law_1.01,0.2519615888595581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,32,power_law_1.2,0.14866559505462645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,32,power_law_1.01,0.26338560581207277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,32,power_law_1.2,0.15039360523223877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,32,power_law_1.01,0.2804095983505249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,32,power_law_1.2,0.1648319959640503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,32,power_law_1.01,0.3080319881439209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,32,power_law_1.2,0.1616320013999939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,32,power_law_1.01,0.35386879444122316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,32,power_law_1.2,0.22716159820556642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,32,power_law_1.01,0.42160639762878416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,32,power_law_1.2,0.22777600288391114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,32,power_law_1.01,0.34783360958099363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,32,power_law_1.01,0.4866687774658203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,32,power_law_1.01,0.42841601371765137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,32,power_law_1.01,0.629369592666626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,32,power_law_1.01,0.5051968097686768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,32,power_law_1.01,0.05436800122261047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,32,power_law_1.01,0.6715456008911133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,32,power_law_1.01,0.05482239723205566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,32,power_law_1.01,0.8507776260375977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,32,power_law_1.01,0.054655998945236206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,32,power_law_1.01,0.055276799201965335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,32,power_law_1.2,0.23888640403747557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,32,power_law_1.01,1.1873344421386718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,32,power_law_1.01,0.055270397663116456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,32,power_law_1.2,0.2551232099533081
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,32,power_law_1.01,0.055852800607681274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,32,power_law_1.2,0.26267518997192385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,32,power_law_1.01,0.05941759943962097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,32,power_law_1.01,1.4781567573547363
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,32,power_law_1.01,0.05980799794197082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,32,power_law_1.01,0.08689280152320862
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,32,power_law_1.01,0.08195840120315552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,32,power_law_1.01,2.171449661254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,32,power_law_1.01,0.07819520235061646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,32,power_law_1.01,0.19137279987335204
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,32,power_law_1.01,0.1819264054298401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,32,power_law_1.2,0.27017600536346437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,32,power_law_1.01,0.16088320016860963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,32,power_law_1.01,2.785753631591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,32,power_law_1.2,0.29748480319976806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,32,power_law_1.01,0.09136639833450318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,32,power_law_1.2,0.08840960264205933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,32,power_law_1.2,0.3722239971160889
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,32,power_law_1.01,0.08570240139961242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,32,power_law_1.2,0.09196799993515015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,32,power_law_1.01,0.08977280259132385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,32,power_law_1.2,0.4135104179382324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,32,power_law_1.2,0.09254400134086609
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,32,power_law_1.01,0.10016640424728393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,32,power_law_1.2,0.09334400296211243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,32,power_law_1.2,0.500761604309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,32,power_law_1.01,0.10748159885406494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,32,power_law_1.2,0.08611199855804444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,32,power_law_1.01,0.1344704031944275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,32,power_law_1.2,0.585804796218872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,32,power_law_1.01,0.789958381652832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,32,power_law_1.2,0.09459199905395507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,32,power_law_1.01,0.1614848017692566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,32,power_law_1.2,0.7865087985992432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,32,power_law_1.2,0.12240639925003052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,32,power_law_1.01,0.21826560497283937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,32,power_law_1.01,1.1118080139160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,32,power_law_1.2,0.11815040111541748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,32,power_law_1.01,0.27200639247894287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,32,power_law_1.2,0.9722432136535645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,32,power_law_1.2,0.1284991979598999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,32,power_law_1.01,1.3747136116027832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,32,power_law_1.01,0.3854207992553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,32,power_law_1.2,0.13004159927368164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,32,power_law_1.2,1.3666111946105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,32,power_law_1.2,0.1860479950904846
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,32,power_law_1.01,0.5070144176483155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,32,power_law_1.2,0.19820799827575683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,32,power_law_1.01,1.9464191436767577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,32,power_law_1.2,0.2006079912185669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,32,power_law_1.01,0.7073472023010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,32,power_law_1.2,1.8037439346313477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,32,power_law_1.2,0.21830399036407472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,32,power_law_1.2,0.2255039930343628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,32,power_law_1.01,0.9683839797973632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,32,power_law_1.01,2.5474367141723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,32,power_law_1.2,0.23917438983917236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,32,power_law_1.2,0.09919999837875366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,32,power_law_1.2,0.2739648103713989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,32,power_law_1.2,2.7121408462524412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,32,power_law_1.2,0.1055616021156311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,32,power_law_1.2,0.31570560932159425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,32,power_law_1.2,0.10597759485244751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,32,power_law_1.2,0.36357119083404543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,32,power_law_1.2,0.10646400451660157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,32,power_law_1.01,3.807379150390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,32,power_law_1.2,0.09287679791450501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,32,power_law_1.2,0.4436800003051758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,32,power_law_1.2,3.3693824768066407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,32,power_law_1.2,0.1044160008430481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,32,power_law_1.2,0.535148811340332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,32,power_law_1.2,0.11463680267333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,32,power_law_1.2,0.11703039407730102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,32,power_law_1.2,0.2140671968460083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,32,power_law_1.2,0.2227776050567627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,32,power_law_1.01,4.9168449401855465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,32,power_law_1.2,0.23018879890441896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,32,power_law_1.2,0.24406399726867675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,32,power_law_1.2,0.2498687982559204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,32,power_law_1.2,0.2518656015396118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,32,power_law_1.2,0.6900352001190185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,32,power_law_1.2,0.29662721157073973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,32,power_law_1.2,0.3261183977127075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,32,power_law_1.2,0.8923520088195801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,32,power_law_1.2,0.3492480039596558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,32,1,balanced,0.03789866715669632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,32,1,balanced,0.04651733239491781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,32,1,balanced,0.062463998794555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,32,power_law_1.2,0.4228799819946289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,32,1,balanced,0.09355733791987102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,32,power_law_1.2,1.226598358154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,32,1,balanced,0.16808533668518066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,32,1,balanced,0.3290560046831767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,32,1,balanced,0.2813760042190552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,32,power_law_1.2,0.5129727840423584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,32,1,balanced,0.279914657274882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,32,1,balanced,0.20612800121307373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,32,1,balanced,0.2079520026842753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,32,1,balanced,0.2584213415781657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,32,power_law_1.2,1.6079488754272462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,32,1,balanced,0.25496532519658405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,32,1,balanced,0.2595679958661397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,32,1,balanced,0.26416534185409546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,32,1,balanced,0.2691199978192647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,32,1,power_law_1.2,0.8787199974060058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,32,1,balanced,0.27481599648793537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,32,1,balanced,0.28657066822052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,32,1,balanced,0.37763198216756183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,32,1,balanced,0.3932373523712158
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,32,power_law_1.2,0.05743359923362732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,32,1,balanced,0.43003201484680176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,32,1,power_law_1.2,0.27388160228729247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,32,1,balanced,0.46937068303426105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,32,power_law_1.2,0.05642240047454834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,32,1,balanced,0.5460906823476156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,32,1,balanced,0.6263733307520548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,32,power_law_1.2,0.05616000294685364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,32,power_law_1.2,0.6515200138092041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,32,1,balanced,0.9552799860636393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,32,power_law_1.2,0.055327999591827395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,32,1,balanced,1.1093653043111165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,32,power_law_1.2,0.05389440059661865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,32,power_law_1.2,0.05775359869003296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,32,1,balanced,1.6019946734110515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,32,power_law_1.2,2.390559959411621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,32,power_law_1.2,0.06325119733810425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,32,power_law_1.2,0.06584960222244263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,32,1,balanced,2.101738611857096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,32,1,balanced,0.03154666721820831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,32,1,balanced,0.03586133321126302
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,32,power_law_1.2,0.08753920197486878
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,32,power_law_1.2,0.08721280097961426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,32,power_law_1.2,3.0332223892211916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,32,power_law_1.2,0.08124799728393554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,32,power_law_1.2,0.8314304351806641
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,32,power_law_1.2,0.1916416049003601
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,32,power_law_1.2,0.17710720300674437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,32,power_law_1.2,1.2434304237365723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,32,power_law_1.2,0.13356800079345704
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,32,power_law_1.2,0.09826560020446777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,32,1,balanced,0.04165866722663244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,32,power_law_1.2,1.494662380218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,32,1,balanced,0.06229333579540253
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,32,power_law_1.2,0.0877568006515503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,32,1,balanced,0.09918933113416036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,32,1,balanced,0.15889066457748413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,32,1,balanced,0.19545066356658936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,32,power_law_1.2,0.09293439984321594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,32,1,balanced,0.19553599754969278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,32,1,balanced,0.18449066082636514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,32,1,balanced,0.1858933369318644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,32,1,balanced,0.22805333137512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,32,1,balanced,0.22907199462254843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,32,power_law_1.2,2.1220672607421873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,32,1,balanced,0.23222400744756064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,32,1,balanced,0.2382240096728007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,32,1,balanced,0.24558399120966592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,32,1,balanced,0.251525342464447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,32,1,balanced,0.2650773326555888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,32,1,balanced,0.3550506830215454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,32,1,balanced,0.37373332182566327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,32,1,balanced,0.41410664717356366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,32,power_law_1.2,2.850111961364746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,32,1,balanced,0.45493332544962567
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,32,1,balanced,0.5390613476435343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,32,1,balanced,0.6236533323923746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,32,power_law_1.2,0.10563839673995971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,32,1,balanced,0.9566240310668945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,32,power_law_1.2,0.11597440242767335
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,32,1,balanced,0.02958400050799052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,32,1,balanced,0.03141333411137263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,32,1,balanced,1.1171092987060547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,32,1,balanced,0.03404266635576884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,32,power_law_1.2,0.14542720317840577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,32,1,balanced,0.041690667470296226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,32,1,balanced,0.05776533484458923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,32,1,balanced,0.09706133604049683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,32,power_law_1.2,4.243417739868164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,32,power_law_1.2,0.16836479902267457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,32,1,balanced,0.10115733742713928
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,32,1,balanced,0.10452799995740254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,32,1,power_law_1.2,1.0539072036743165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,32,power_law_1.2,0.2269887924194336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,32,1,power_law_1.01,0.03776000142097473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,32,power_law_1.2,0.29845759868621824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,32,1,power_law_1.01,0.044870400428771974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,32,1,power_law_1.01,0.06262400150299072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,32,power_law_1.2,0.41491198539733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,32,1,power_law_1.01,0.09304320216178893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,32,power_law_1.2,0.5260672092437744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,32,1,power_law_1.01,0.13742079734802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,32,power_law_1.2,5.685190582275391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,32,1,power_law_1.01,0.18360320329666138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,32,1,balanced,1.6137812932332356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,32,power_law_1.2,0.8691647529602051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,32,1,balanced,0.10729066530863444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,32,1,power_law_1.01,0.02985599935054779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,32,1,power_law_1.01,0.21671040058135987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,32,1,power_law_1.01,0.033452799916267394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,32,1,power_law_1.01,0.22447359561920166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,32,1,balanced,2.1105333964029946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,32,power_law_1.2,1.1609600067138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,32,1,power_law_1.01,0.046028798818588255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,32,1,power_law_1.01,0.1787392020225525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,32,1,power_law_1.2,0.03790720105171204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,32,1,power_law_1.01,0.0610368013381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,32,1,power_law_1.01,0.18279039859771729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,32,1,power_law_1.2,0.04479359984397888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,32,1,power_law_1.01,0.08388479948043823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,32,1,power_law_1.01,0.23752319812774658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,32,1,power_law_1.2,0.06297600269317627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,32,1,power_law_1.01,0.10767359733581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,32,1,power_law_1.01,0.2371840000152588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,32,1,power_law_1.2,0.09226880073547364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,32,1,power_law_1.01,0.15543680191040038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,32,1,power_law_1.01,0.24216959476470948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,32,1,balanced,0.10714667042096455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,32,1,power_law_1.2,0.12803839445114135
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,32,1,balanced,0.11195733149846394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,32,1,power_law_1.01,0.15942399501800536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,32,1,power_law_1.01,0.24544639587402345
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,32,1,balanced,0.11689600348472595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,32,1,balanced,0.12359999616940816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,32,1,power_law_1.2,0.16940799951553345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,32,1,power_law_1.01,0.1586303949356079
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,32,1,balanced,0.13136000434557596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,32,1,balanced,0.16018666823705038
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,32,1,balanced,0.16981865962346396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,32,1,power_law_1.01,0.16437120437622071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,32,1,power_law_1.2,0.21317760944366454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,32,1,balanced,0.18946133057276407
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,32,1,balanced,0.19222933053970337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,32,1,power_law_1.01,0.20801279544830323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,32,1,balanced,0.21786133448282877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,32,1,power_law_1.2,0.2200256109237671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,32,1,power_law_1.2,0.6400000095367432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,32,1,balanced,0.2424373428026835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,32,1,power_law_1.01,0.21196799278259276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,32,1,balanced,0.27853333950042725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,32,1,balanced,0.4031733274459839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,32,1,power_law_1.01,0.2162559986114502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,32,1,balanced,0.4736959934234619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,32,1,power_law_1.01,0.22458240985870362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,32,1,balanced,0.6298559904098511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,32,1,power_law_1.01,0.23023359775543212
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,32,1,balanced,0.6856799920399984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,32,1,power_law_1.01,0.25520639419555663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,32,1,power_law_1.01,0.2413952112197876
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,32,1,balanced,0.9801599979400635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,32,1,power_law_1.01,0.2563071966171265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,32,1,power_law_1.01,0.3449984073638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,32,1,balanced,1.2750879923502605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,32,1,power_law_1.01,0.028339201211929323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,32,1,power_law_1.2,0.1775936007499695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,32,1,power_law_1.01,0.3740288019180298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,32,1,power_law_1.2,0.030348798632621764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,32,1,power_law_1.01,0.029631999135017396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,32,1,power_law_1.2,0.18136320114135743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,32,1,power_law_1.2,0.03331199884414673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,32,1,power_law_1.01,0.03272959887981415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,32,1,power_law_1.2,0.23590400218963622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,32,1,power_law_1.2,0.04664320051670075
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,32,1,power_law_1.01,0.04069119989871979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,32,1,power_law_1.2,0.23394560813903809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,32,1,power_law_1.2,0.06069120168685913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,32,1,power_law_1.2,0.2405695915222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,32,1,power_law_1.01,0.05309439897537231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,32,1,power_law_1.2,0.07900159955024719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,32,1,power_law_1.01,0.2644160032272339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,32,1,power_law_1.2,0.25014400482177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,32,1,power_law_1.01,0.07502719759941101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,32,1,power_law_1.2,0.10054399967193603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,32,1,power_law_1.01,0.280134391784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,32,1,power_law_1.2,0.25505919456481935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,32,1,power_law_1.01,0.09068160057067871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,32,1,power_law_1.2,0.15456639528274535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,32,1,power_law_1.01,0.09531520009040832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,32,1,power_law_1.2,0.15872639417648315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,32,1,power_law_1.01,0.4280384063720703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,32,1,power_law_1.01,0.09839360117912292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,32,1,power_law_1.2,0.15806080102920533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,32,1,power_law_1.01,0.5127295970916748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,32,1,power_law_1.01,0.10126719474792481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,32,1,power_law_1.2,0.16200319528579712
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,32,1,power_law_1.01,0.10744320154190064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,32,1,power_law_1.01,0.6253568172454834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,32,1,power_law_1.2,0.20780160427093505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,32,1,power_law_1.01,0.11685760021209717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,32,1,power_law_1.2,0.21164159774780272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,32,1,power_law_1.01,0.7454016208648682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,32,1,power_law_1.01,0.12688640356063843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,32,1,power_law_1.2,0.2194751977920532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,32,1,power_law_1.01,0.13595520257949828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,32,1,power_law_1.01,0.36119039058685304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,32,1,power_law_1.2,0.22510080337524413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,32,1,power_law_1.01,0.9870976448059082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,32,1,power_law_1.01,0.15348479747772217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,32,1,power_law_1.01,0.3944384098052979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,32,1,power_law_1.2,0.23535358905792236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,32,1,power_law_1.01,0.16684160232543946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,32,1,power_law_1.2,0.2458048105239868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,32,1,power_law_1.01,0.44744319915771485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,32,1,power_law_1.01,1.2350079536437988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,32,1,power_law_1.01,0.19536639451980592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,32,1,power_law_1.2,0.26335361003875735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,32,1,power_law_1.01,0.5466752052307129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,32,1,power_law_1.01,0.18908159732818602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,32,1,power_law_1.2,0.3578239917755127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,32,1,power_law_1.01,0.22544000148773194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,32,1,power_law_1.01,0.6580160140991211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,32,1,power_law_1.2,0.387174391746521
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,32,1,power_law_1.01,0.2416640043258667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,32,1,power_law_1.2,0.4707968235015869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,32,1,power_law_1.2,0.2670016050338745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,32,1,power_law_1.01,0.7659264087677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,32,1,power_law_1.01,0.2899967908859253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,32,1,power_law_1.2,0.2854399919509888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,32,1,power_law_1.2,0.525267219543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,32,1,power_law_1.01,0.4145472049713135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,32,1,power_law_1.01,0.9913663864135742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,32,1,power_law_1.2,0.3723135948181152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,32,1,power_law_1.2,0.6431551933288574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,32,1,power_law_1.01,0.5536255836486816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,32,1,power_law_1.2,0.4033088207244873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,32,1,power_law_1.01,1.25098876953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,32,1,power_law_1.2,0.7618815898895264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,32,1,power_law_1.01,0.5850111961364746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,32,1,power_law_1.2,0.5080383777618408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,32,1,power_law_1.01,1.7335487365722657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,32,1,power_law_1.2,0.9993727684020997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,32,1,power_law_1.01,0.7364543914794922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,32,1,power_law_1.2,0.5614848136901855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,32,1,power_law_1.01,1.7458560943603516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,32,1,power_law_1.2,0.6679168224334717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,32,1,power_law_1.2,1.2558079719543458
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,32,1,power_law_1.01,1.0254207611083985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,32,1,power_law_1.01,2.2339391708374023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,32,1,power_law_1.2,0.7768447875976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,32,1,power_law_1.01,2.245721626281738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,32,balanced,0.04348266621430715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,32,1,power_law_1.01,1.3462464332580566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,32,balanced,0.0436106671889623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,32,1,power_law_1.2,1.7467199325561524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,32,balanced,0.03781333317359289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,32,balanced,0.03893866638342539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,32,balanced,0.0378560001651446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,32,balanced,0.04811733464399973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,32,balanced,0.06517333288987477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,32,balanced,0.09431999921798706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,32,balanced,0.11682666341463725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,32,balanced,0.11777066191037495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,32,balanced,0.122597336769104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,32,balanced,0.12369599938392639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,32,balanced,0.17111466328303018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,32,1,power_law_1.2,2.2532800674438476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,32,balanced,0.18133866786956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,32,balanced,0.18525334199269614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,32,balanced,0.19718400637308756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,32,balanced,0.05620799958705902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,32,balanced,0.05416533350944519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,32,balanced,0.20728532473246256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,32,balanced,0.05418666700522105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,32,balanced,0.21614933013916016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,32,balanced,0.06402666866779327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,32,balanced,0.08060800035794576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,32,balanced,0.2363626758257548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,32,balanced,0.10823466380437215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,32,balanced,0.28061334292093915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,32,balanced,0.10931733250617981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,32,balanced,0.31200534105300903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,32,balanced,0.04367466767628988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,32,balanced,0.37109867731730145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,32,balanced,0.06047999858856201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,32,1,power_law_1.2,1.0042176246643066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,32,balanced,0.08383466800053914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,32,balanced,0.43406399091084796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,32,balanced,0.15103466312090555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,32,balanced,0.14087999860445657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,32,balanced,0.5560959974924723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,32,balanced,0.14258133371671042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,32,balanced,0.15200533469518027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,32,balanced,0.15189866224924722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,32,balanced,0.6806933085123698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,32,balanced,0.19784533977508545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,32,1,power_law_1.2,1.2607232093811036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,32,balanced,0.20324800411860147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,32,balanced,0.20775467157363892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,32,balanced,0.9616853396097819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,32,balanced,0.21755200624465942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,32,balanced,0.223797341187795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,32,balanced,0.23435733715693155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,32,balanced,1.20906662940979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,32,1,power_law_1.2,1.7619840621948242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,32,balanced,0.11005333065986633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,32,1,power_law_1.2,0.3537408113479614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,32,1,power_law_1.2,0.026707199215888978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,32,balanced,0.21270400285720825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,32,balanced,1.7359573046366374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,32,balanced,0.21400533119837442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,32,balanced,0.2216213345527649
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,32,1,power_law_1.2,0.028415998816490172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,32,balanced,0.22864532470703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,32,balanced,0.23430399099985758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,32,1,power_law_1.2,0.031737598776817325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,32,1,power_law_1.2,2.2669567108154296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,32,balanced,2.2573599815368652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,32,1,power_law_1.2,0.03845759928226471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,32,power_law_1.01,0.1326591968536377
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,32,1,power_law_1.2,0.04990079998970032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,32,power_law_1.01,0.13111040592193604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,32,balanced,0.2493013342221578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,32,1,power_law_1.2,0.07208319902420043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,32,power_law_1.01,0.13203840255737304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,32,1,power_law_1.2,0.08865919709205627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,32,power_law_1.01,0.13389439582824708
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,32,1,power_law_1.2,0.09308159947395325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,32,power_law_1.01,0.13152639865875243
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,32,1,power_law_1.2,0.09626880288124084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,32,balanced,0.2489173412322998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,32,power_law_1.01,0.13722239732742308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,32,balanced,0.2608853379885356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,32,1,power_law_1.2,0.10028159618377686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,32,balanced,0.2731786568959554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,32,power_law_1.01,0.1328703999519348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,32,balanced,0.29917333523432416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,32,1,power_law_1.2,0.10609279870986939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,32,balanced,0.3447626829147339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,32,power_law_1.01,0.12833919525146484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,32,1,power_law_1.2,0.11527680158615113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,32,balanced,0.39025600751241046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,32,power_law_1.01,0.14237439632415771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,32,balanced,0.4842880169550578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,32,balanced,0.3141439954439799
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,32,1,power_law_1.2,0.12578560113906861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,32,balanced,0.3438719908396403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,32,power_law_1.01,0.1468160033226013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,32,balanced,0.5764266649881998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,32,1,power_law_1.2,0.1373695969581604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,32,balanced,0.396778662999471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,32,balanced,0.4533173243204753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,32,balanced,0.8224960168202718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,32,1,power_law_1.2,0.15418239831924438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,32,balanced,0.5643839836120605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,32,balanced,1.0094506740570068
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,32,1,power_law_1.2,0.16637439727783204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,32,balanced,0.6752800146738688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,32,1,power_law_1.2,0.1958016037940979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,32,balanced,0.9604907035827637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,32,balanced,1.4411360422770183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,32,1,power_law_1.2,0.19022719860076903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,32,balanced,1.1855200131734211
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,32,1,power_law_1.2,0.22547199726104736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,32,balanced,1.8742879231770833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,32,1,power_law_1.2,0.24246399402618407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,32,balanced,1.7011094093322754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,32,power_law_1.01,0.1814784049987793
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,32,1,power_law_1.2,0.2904383897781372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,32,balanced,2.7309707005818686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,32,power_law_1.01,0.19741439819335938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,32,balanced,2.216485341389974
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,32,1,power_law_1.2,0.4752064228057861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,32,power_law_1.01,0.20154879093170167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,32,power_law_1.01,0.07914239764213563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,32,1,power_law_1.2,0.5167808055877685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,32,power_law_1.01,0.20940160751342773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,32,power_law_1.01,0.08183040022850037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,32,power_law_1.01,0.2170367956161499
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,32,1,power_law_1.2,0.5881087779998779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,32,power_law_1.01,0.08369920253753663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,32,power_law_1.01,0.24040958881378174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,32,power_law_1.01,0.08271359801292419
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,32,balanced,0.029882666965325672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,32,power_law_1.01,0.2617023944854736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,32,1,power_law_1.2,0.7416255950927735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,32,power_law_1.01,0.08065919876098633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,32,power_law_1.01,0.32079999446868895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,32,power_law_1.01,0.08155519962310791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,32,power_law_1.01,0.35656960010528566
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,32,1,power_law_1.2,1.0282624244689942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,32,power_law_1.01,0.1032256007194519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,32,power_law_1.01,0.42947840690612793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,32,power_law_1.01,0.11140480041503906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,32,power_law_1.01,0.4974976062774658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,32,power_law_1.01,0.10780160427093506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,32,1,power_law_1.2,1.3648192405700683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,32,balanced,3.590858777364095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,32,power_law_1.01,0.11317119598388672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,32,power_law_1.01,0.6523712158203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,32,power_law_1.01,0.08840320110321045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,32,power_law_1.01,0.16162559986114503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,32,balanced,0.03031466652949651
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,32,balanced,0.029738667110602062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,32,power_law_1.01,0.09476479887962341
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,32,balanced,0.03181333343187968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,32,power_law_1.01,0.8035455703735351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,32,power_law_1.01,0.178656005859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,32,balanced,0.03770666569471359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,32,balanced,0.05886933207511902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,32,balanced,0.06230400005976359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,32,power_law_1.01,0.09489920139312744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,32,balanced,0.06449066599210103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,32,balanced,0.08775466680526733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,32,balanced,0.09130133191744487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,32,balanced,0.08524266878763835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,32,balanced,0.2005280057589213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,32,balanced,0.1944053371747335
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,32,balanced,0.16151466965675354
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,32,balanced,0.12359467148780823
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,32,balanced,0.08240533371766408
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,32,balanced,0.08855467041333516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,32,balanced,0.09708266456921895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,32,balanced,0.10341866811116536
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,32,balanced,0.11812266707420349
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,32,balanced,0.13582932949066162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,32,balanced,0.1707893411318461
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,32,balanced,0.20823466777801514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,32,power_law_1.01,0.1729599952697754
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,32,balanced,0.29341866572697956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,32,power_law_1.01,0.18721280097961426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,32,power_law_1.01,0.09578880071640014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,32,balanced,0.3676266670227051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,32,power_law_1.01,0.20035200119018554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,32,power_law_1.01,0.09036160111427308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,32,balanced,0.5199146668116251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,32,power_law_1.01,0.21671040058135987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,32,power_law_1.01,0.10400639772415161
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,32,balanced,0.6785866419474283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,32,power_law_1.01,0.24128639698028564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,32,power_law_1.01,0.103603196144104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,32,power_law_1.2,0.13156479597091675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,32,power_law_1.01,0.285696005821228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,32,power_law_1.01,0.10584959983825684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,32,power_law_1.2,0.13035520315170288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,32,power_law_1.01,0.3223743915557861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,32,power_law_1.01,0.1978432059288025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,32,power_law_1.2,0.13317760229110717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,32,power_law_1.01,0.3909503936767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,32,power_law_1.01,0.2089087963104248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,32,power_law_1.2,0.13144960403442382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,32,power_law_1.01,1.1292351722717284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,32,power_law_1.01,0.21758720874786378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,32,power_law_1.2,0.1142848014831543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,32,power_law_1.01,0.21679999828338622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,32,power_law_1.2,0.13319679498672485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,32,power_law_1.01,1.3985535621643066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,32,power_law_1.01,0.22392959594726564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,32,power_law_1.2,0.12906880378723146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,32,power_law_1.01,0.24607360363006592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,32,power_law_1.2,0.1364799976348877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,32,power_law_1.01,0.25504000186920167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,32,power_law_1.01,2.078764724731445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,32,power_law_1.2,0.14338560104370118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,32,power_law_1.01,0.2667327880859375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,32,power_law_1.2,0.14392319917678834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,32,power_law_1.01,0.30995199680328367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,32,power_law_1.2,0.18405760526657106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,32,power_law_1.01,0.374451208114624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,32,power_law_1.2,0.1866752028465271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,32,power_law_1.01,0.43902077674865725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,32,power_law_1.2,0.20113279819488525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,32,power_law_1.01,2.741708755493164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,32,power_law_1.2,0.20032000541687012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,32,power_law_1.01,0.5593279838562012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,32,power_law_1.2,0.23398399353027344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,32,power_law_1.01,0.7159679889678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,32,power_law_1.2,0.24625918865203858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,32,power_law_1.2,0.26705920696258545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,32,power_law_1.01,0.9744895935058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,32,power_law_1.2,0.3287935972213745
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,32,power_law_1.01,0.05192959904670715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,32,power_law_1.2,0.36732161045074463
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,32,power_law_1.01,0.050963199138641356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,32,power_law_1.2,0.45369601249694824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,32,power_law_1.01,0.05146880149841308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,32,power_law_1.2,0.07948160171508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,32,power_law_1.2,0.519814395904541
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,32,power_law_1.01,0.05160319805145264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,32,power_law_1.2,0.08230400085449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,32,power_law_1.2,0.08403199911117554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,32,power_law_1.01,0.05027840137481689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,32,power_law_1.2,0.6884352207183838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,32,power_law_1.2,0.08361600041389465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,32,power_law_1.01,0.05075200200080872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,32,1,power_law_1.2,0.8248703956604004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,32,power_law_1.2,0.8535103797912598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,32,power_law_1.01,0.46888318061828616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,32,power_law_1.2,0.0770687997341156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,32,power_law_1.01,0.05726079940795899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,32,power_law_1.01,1.2120832443237304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,32,power_law_1.01,0.05931519865989685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,32,power_law_1.2,0.09054719805717468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,32,power_law_1.2,1.27325439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,32,power_law_1.2,0.09445120096206665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,32,power_law_1.2,0.09590399861335755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,32,power_law_1.01,1.7610944747924804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,32,power_law_1.2,1.5716927528381348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,32,power_law_1.2,0.09679359793663025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,32,1,power_law_1.2,1.2679360389709473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,32,power_law_1.2,0.09227520227432251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,32,power_law_1.2,0.09715840220451355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,32,power_law_1.01,2.3484159469604493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,32,power_law_1.2,2.2567615509033203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,32,power_law_1.01,0.6139328002929687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,32,power_law_1.2,0.10373760461807251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,32,power_law_1.01,0.08230400085449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,32,power_law_1.2,0.10915199518203736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,32,power_law_1.01,0.7501696109771728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,32,power_law_1.01,0.07464960217475891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,32,power_law_1.2,0.19367040395736695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,32,power_law_1.01,0.07559679746627808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,32,power_law_1.2,0.20862081050872802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,32,power_law_1.01,3.480723190307617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,32,power_law_1.2,3.279916763305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,32,power_law_1.01,0.14935040473937988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,32,power_law_1.2,0.21823999881744385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,32,power_law_1.2,0.21626880168914794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,32,power_law_1.01,0.15765759944915772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,32,power_law_1.2,0.21923201084136962
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,32,power_law_1.01,0.12614400386810304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,32,power_law_1.2,0.08206719756126404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,32,power_law_1.2,0.24360320568084717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,32,power_law_1.01,0.0993664026260376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,32,power_law_1.2,0.10360959768295289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,32,power_law_1.01,0.07728639841079712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,32,power_law_1.2,0.2704319953918457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,32,power_law_1.2,0.10814720392227173
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,32,power_law_1.01,0.08250240087509156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,32,power_law_1.2,0.2811840057373047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,32,power_law_1.2,0.10366719961166382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,32,power_law_1.01,1.060934352874756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,32,power_law_1.01,0.09044479727745056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,32,power_law_1.2,0.3151616096496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,32,power_law_1.2,0.11302399635314941
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,32,power_law_1.01,0.10147199630737305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,32,power_law_1.2,0.1590208053588867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,32,power_law_1.2,0.38471040725708006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,32,1,power_law_1.2,0.5018943786621094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,32,power_law_1.01,1.344761562347412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,32,power_law_1.2,0.1737215995788574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,32,power_law_1.01,0.12276480197906495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,32,power_law_1.2,0.44644479751586913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,32,power_law_1.2,0.17820800542831422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,32,power_law_1.01,0.1453503966331482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,32,power_law_1.2,0.1907520055770874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,32,power_law_1.01,0.18848639726638794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,32,power_law_1.2,0.6149824142456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,32,power_law_1.01,1.9523839950561523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,32,power_law_1.2,0.20448000431060792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,32,power_law_1.01,0.24581758975982665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,32,power_law_1.01,4.452902221679688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,32,power_law_1.2,0.21957120895385743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,32,power_law_1.2,0.7643392086029053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,32,power_law_1.01,0.34025599956512453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,32,power_law_1.2,0.2477247953414917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,32,power_law_1.2,0.292576003074646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,32,power_law_1.01,0.44420480728149414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,32,power_law_1.01,2.5973119735717773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,32,power_law_1.2,0.3272511959075928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,32,1,balanced,0.03374933451414108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,32,1,balanced,0.04159999887148539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,32,1,balanced,0.05477866530418396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,32,1,balanced,0.08265066643555959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,32,1,balanced,0.1466506620248159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,32,power_law_1.2,0.40119037628173826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,32,1,balanced,0.2839306592941284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,32,1,balanced,0.2448479930559794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,32,1,balanced,0.24587732553482056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,32,power_law_1.2,0.475494384765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,32,power_law_1.2,0.05204480290412903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,32,1,balanced,0.1811199982961019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,32,1,balanced,0.1827253301938375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,32,1,balanced,0.2250773310661316
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,32,power_law_1.2,0.05289599895477295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,32,power_law_1.2,0.6292479991912842
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,32,power_law_1.2,0.052108800411224364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,32,power_law_1.2,0.7962111949920654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,32,power_law_1.2,1.0242239952087402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,32,power_law_1.2,0.05244799852371216
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,32,power_law_1.2,0.05117440223693848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,32,power_law_1.2,1.1198975563049316
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,32,power_law_1.01,0.6238272190093994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,32,power_law_1.2,1.341254425048828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,32,power_law_1.2,0.04943360090255737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,32,1,balanced,0.029792000850041706
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,32,power_law_1.2,0.05568000078201294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,32,1,balanced,0.03357866654793421
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,32,power_law_1.01,0.8711423873901367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,32,1,balanced,0.03742400060097376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,32,power_law_1.2,1.409273624420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,32,1,balanced,0.05189866820971171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,32,1,balanced,0.08899733424186707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,32,1,balanced,0.22232532501220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,32,power_law_1.2,0.060140800476074216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,32,1,balanced,0.22432533899943033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,32,power_law_1.2,1.8987392425537108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,32,1,balanced,0.22984000047047934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,32,1,balanced,0.23409066597620645
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,32,power_law_1.2,0.07631999850273133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,32,1,balanced,0.23875733216603598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,32,1,balanced,0.24980266888936362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,32,power_law_1.2,0.07779200077056884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,32,1,balanced,0.32796265681584674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,32,power_law_1.2,2.06375675201416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,32,1,balanced,0.34273600578308105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,32,1,balanced,0.3755413293838501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,32,1,balanced,0.4086986780166626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,32,power_law_1.2,2.5230335235595702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,32,1,balanced,0.4752906560897827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,32,1,balanced,0.5437546571095785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,32,power_law_1.2,2.6101119995117186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,32,1,balanced,0.8306453227996826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,32,1,balanced,0.1414240002632141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,32,1,power_law_1.01,0.0331712007522583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,32,1,balanced,0.17114667097727457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,32,1,balanced,0.9669600327809652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,32,1,balanced,0.1718453367551168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,32,1,balanced,0.16106667121251425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,32,1,power_law_1.01,0.03985919952392578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,32,1,balanced,0.16364799936612448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,32,1,balanced,0.19979200760523477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,32,1,balanced,1.3975520133972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,32,power_law_1.2,3.7021312713623047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,32,1,power_law_1.01,0.05479679703712463
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,32,power_law_1.2,0.07406719923019409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,32,1,power_law_1.01,0.08124799728393554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,32,1,balanced,1.8326773643493652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,32,power_law_1.2,0.16625280380249025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,32,1,power_law_1.01,0.11838079690933227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,32,1,power_law_1.01,0.029216000437736513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,32,power_law_1.2,0.15695359706878662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,32,1,power_law_1.01,0.16043519973754883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,32,1,power_law_1.01,0.031667199730873105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,32,1,balanced,0.029045333464940388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,32,power_law_1.2,0.1286463975906372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,32,1,balanced,0.02941333254178365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,32,1,power_law_1.01,0.18731520175933838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,32,1,balanced,0.03315199911594391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,32,1,power_law_1.01,0.04190720021724701
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,32,1,balanced,0.03895466774702072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,32,1,balanced,0.051882664362589516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,32,power_law_1.2,0.10142079591751099
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,32,1,balanced,0.08715732892354329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,32,1,balanced,0.08904533584912618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,32,1,power_law_1.01,0.0553600013256073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,32,power_law_1.2,0.07787520289421082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,32,power_law_1.2,4.998963165283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,32,1,balanced,0.20142932732899985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,32,1,power_law_1.01,0.07461119890213012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,32,power_law_1.2,0.08190079927444457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,32,1,balanced,0.20360533396402994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,32,1,balanced,0.2107306718826294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,32,1,balanced,0.21586666504542032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,32,1,power_law_1.01,0.09555839896202087
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,32,power_law_1.2,0.09631360173225403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,32,1,balanced,0.22318933407465616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,32,1,balanced,0.23216533660888672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,32,1,power_law_1.01,0.1377344012260437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,32,1,balanced,0.31335999568303424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,32,power_law_1.2,0.10523519515991211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,32,1,balanced,0.32889066139856976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,32,1,power_law_1.01,0.14257919788360596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,32,power_law_1.2,0.12692480087280272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,32,1,power_law_1.01,0.1424831986427307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,32,power_law_1.2,0.15569280385971068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,32,1,power_law_1.01,0.14264320135116576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,32,1,power_law_1.01,0.19747840166091918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,32,power_law_1.2,0.21075201034545898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,32,1,balanced,0.09390399853388469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,32,1,power_law_1.01,0.18235520124435425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,32,1,balanced,0.09486933549245198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,32,1,power_law_1.01,0.15678720474243163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,32,1,balanced,0.09669333696365356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,32,power_law_1.2,0.2475584030151367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,32,1,balanced,0.10090667009353638
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,32,1,balanced,0.10359999537467957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,32,1,power_law_1.01,0.15865600109100342
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,32,1,balanced,0.1079360047976176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,32,1,balanced,0.11735999584197998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,32,power_law_1.2,0.385100793838501
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,32,1,balanced,0.1421173314253489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,32,1,power_law_1.01,0.20517120361328126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,32,1,balanced,0.14877333243687949
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,32,1,balanced,0.16710933049519858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,32,1,power_law_1.01,0.20662400722503663
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,32,1,balanced,0.16889599959055582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,32,power_law_1.2,0.463212776184082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,32,1,balanced,0.3632213274637858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,32,1,balanced,0.19161067406336466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,32,1,power_law_1.01,0.20942718982696534
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,32,1,balanced,0.21229867140452066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,32,1,balanced,0.24317334095637003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,32,1,power_law_1.01,0.21631999015808107
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,32,power_law_1.2,0.7187776088714599
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,32,1,balanced,0.3485226631164551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,32,1,power_law_1.01,0.22206079959869385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,32,1,balanced,0.4087093273798625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,32,1,balanced,0.5565173228581747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,32,1,power_law_1.01,0.22849280834198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,32,1,power_law_1.01,0.18757760524749756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,32,power_law_1.2,0.9614399909973145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,32,1,balanced,0.6008586486180624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,32,1,power_law_1.01,0.24277119636535643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,32,1,power_law_1.01,0.19133440256118775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,32,1,power_law_1.2,0.03351039886474609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,32,1,power_law_1.01,0.317740797996521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,32,1,power_law_1.01,0.19782400131225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,32,1,power_law_1.2,0.03953279852867127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,32,1,power_law_1.01,0.34011518955230713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,32,1,power_law_1.01,0.20318078994750977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,32,1,power_law_1.2,0.055225598812103274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,32,1,balanced,0.39962132771809894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,32,1,power_law_1.01,0.21063039302825928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,32,1,power_law_1.01,0.3884608030319214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,32,1,power_law_1.2,0.08115839958190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,32,1,balanced,0.4732746680577596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,32,1,power_law_1.01,0.22584960460662842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,32,1,balanced,0.547813336054484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,32,1,power_law_1.01,0.47830400466918943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,32,1,power_law_1.01,0.3027328014373779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,32,1,balanced,0.836026668548584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,32,1,power_law_1.01,0.5695936203002929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,32,1,power_law_1.01,0.33098878860473635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,32,1,balanced,0.9810400009155273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,32,1,power_law_1.01,0.02808319926261902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,32,1,power_law_1.01,0.37763199806213377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,32,1,power_law_1.01,0.6642687797546387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,32,1,balanced,1.4187253316243489
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,32,1,power_law_1.01,0.029382398724555968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,32,1,power_law_1.01,0.44768638610839845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,32,1,power_law_1.01,0.8607680320739746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,32,1,power_law_1.01,0.03200640082359314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,32,1,power_law_1.01,0.5508927822113037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,32,1,balanced,1.8507520357767742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,32,1,power_law_1.01,0.03839359879493713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,32,1,power_law_1.2,0.027993598580360414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,32,1,power_law_1.01,0.6532671928405762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,32,1,power_law_1.2,0.11077760457992554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,32,1,power_law_1.01,0.04649600088596344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,32,1,power_law_1.2,0.031200000643730165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,32,1,power_law_1.2,0.1468224048614502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,32,1,power_law_1.01,0.06828160285949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,32,1,power_law_1.01,0.8615936279296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,32,1,power_law_1.2,0.18644479513168336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,32,1,power_law_1.01,0.0797760009765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,32,1,power_law_1.01,1.0828160285949706
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,32,1,power_law_1.01,0.08572800159454345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,32,1,power_law_1.2,0.19575680494308473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,32,1,power_law_1.2,0.15382399559020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,32,1,power_law_1.01,0.08781440258026123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,32,1,power_law_1.01,0.08997759819030762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,32,1,power_law_1.2,1.6640127182006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,32,1,power_law_1.01,1.5197440147399903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,32,1,power_law_1.01,1.0888192176818847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,32,1,power_law_1.01,0.09478399753570557
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,32,1,power_law_1.01,0.10241279602050782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,32,1,power_law_1.2,0.04160000085830688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,32,1,power_law_1.01,0.1106943964958191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,32,1,power_law_1.01,1.5195648193359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,32,1,power_law_1.01,1.959891128540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,32,1,power_law_1.2,0.05411199927330017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,32,balanced,0.029258665939172108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,32,1,power_law_1.01,0.12149120569229126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,32,balanced,0.02934933453798294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,32,balanced,0.029114666084448498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,32,balanced,0.029359998802344005
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,32,1,balanced,0.8477813402811686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,32,balanced,0.031199999153614044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,32,balanced,0.03164800008138021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,32,balanced,0.037402667105197906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,32,balanced,0.037503999968369804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,32,balanced,0.03765333443880081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,32,balanced,0.03766400118668874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,32,1,balanced,1.0986826419830322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,32,balanced,0.045909335215886436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,32,balanced,0.046336000164349876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,32,balanced,0.02739733209212621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,32,balanced,0.04791999856630961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,32,balanced,0.027530667682488758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,32,balanced,0.04985600213209788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,32,balanced,0.05197333296140035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,32,balanced,0.056015998125076294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,32,balanced,0.06465066472689311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,32,balanced,0.07865599791208903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,32,balanced,0.09211732943852742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,32,balanced,0.11447999874750774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,32,balanced,0.13524799545605978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,32,balanced,0.18872000773747763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,32,balanced,0.23427200317382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,32,balanced,0.3287893335024516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,32,balanced,0.4250719944636027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,32,1,power_law_1.2,0.06993280053138733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,32,balanced,0.6158239841461182
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,32,1,power_law_1.01,0.13523839712142943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,32,balanced,0.8035146395365397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,32,1,power_law_1.2,0.09242240190505982
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,32,1,power_law_1.01,0.14522240161895753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,32,1,power_law_1.01,1.9534847259521484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,32,1,power_law_1.2,0.15601279735565185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,32,balanced,0.02959466725587845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,32,1,power_law_1.01,0.17107199430465697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,32,balanced,0.027445333699385326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,32,1,power_law_1.2,0.026099199056625368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,32,balanced,0.029824001093705494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,32,balanced,0.031557333966096245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,32,balanced,0.035536001125971474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,32,1,power_law_1.01,0.16618880033493041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,32,balanced,0.035760000348091125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,32,balanced,0.03754133234421412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,32,balanced,0.03922666609287262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,32,balanced,0.04598933458328247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,32,1,power_law_1.01,0.19631999731063843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,32,balanced,0.04996266464392344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,32,balanced,0.050986667474110924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,32,balanced,0.05197333296140035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,32,balanced,0.05809600154558817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,32,1,power_law_1.01,0.20940160751342773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,32,balanced,0.062165334820747375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,32,balanced,0.06865066786607106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,32,balanced,0.07867200175921123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,32,balanced,0.09247466921806335
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,32,1,power_law_1.01,0.2528575897216797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,32,balanced,0.12366400162378947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,32,balanced,0.14833600322405496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,32,balanced,0.208186666170756
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,32,balanced,0.024154665569464367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,32,balanced,0.02401600033044815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,32,balanced,0.2646613319714864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,32,balanced,0.03753600021203359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,32,balanced,0.023397333920001984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,32,balanced,0.03740799923737844
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,32,balanced,0.025429333249727886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,32,balanced,0.0374293327331543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,32,balanced,0.3716426690419515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,32,balanced,0.03792533278465271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,32,balanced,0.025461333493391674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,32,1,power_law_1.2,0.13552639484405518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,32,balanced,0.03763733307520548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,32,balanced,0.029482667644818623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,32,balanced,0.03952533255020777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,32,balanced,0.0295413335164388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,32,balanced,0.03974399964014689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,32,1,power_law_1.2,0.20499839782714843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,32,balanced,0.03274133304754893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,32,balanced,0.040922666589419045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,32,1,power_law_1.2,0.14056320190429689
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,32,balanced,0.037871999045213066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,32,balanced,0.05443733433882395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,32,balanced,0.0377813329299291
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,32,1,power_law_1.2,0.027558401226997375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,32,1,power_law_1.2,0.20562560558319093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,32,balanced,0.03782933453718821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,32,balanced,0.045968001087506614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,32,balanced,0.04582933088143667
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,32,1,power_law_1.2,0.03134720027446747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,32,balanced,0.03766933331886927
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,32,balanced,0.03839466720819473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,32,balanced,0.037733333806196846
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,32,balanced,0.03861333429813385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,32,1,power_law_1.2,0.035129600763320924
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,32,balanced,0.04190400242805481
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,32,balanced,0.04394666850566864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,32,balanced,0.04975999891757965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,32,balanced,0.05406400064627329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,32,1,power_law_1.2,0.04450559914112091
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,32,balanced,0.06565866867701213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,32,1,power_law_1.01,0.35784320831298827
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,32,1,power_law_1.2,0.06405760049819946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,32,1,power_law_1.01,0.44292478561401366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,32,1,power_law_1.2,0.07682560086250305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,32,balanced,0.054117331902186074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,32,balanced,0.05526400109132131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,32,balanced,0.05733866492907206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,32,balanced,0.05753066639105479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,32,balanced,0.06275199850400288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,32,1,power_law_1.2,0.20918400287628175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,32,balanced,0.06615466872851054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,32,balanced,0.0721919983625412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,32,balanced,0.0812960018714269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,32,1,power_law_1.2,0.21779201030731202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,32,balanced,0.10201066732406616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,32,balanced,0.12139733632405598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,32,balanced,0.15834133823712668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,32,1,power_law_1.2,0.2233151912689209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,32,balanced,0.1954186757405599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,32,balanced,0.2757493257522583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,32,1,power_law_1.2,0.23371520042419433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,32,balanced,0.08062933385372162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,32,balanced,0.351034681002299
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,32,balanced,0.10934933026631673
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,32,balanced,0.1421119968096415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,32,1,power_law_1.2,0.24570879936218262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,32,balanced,0.5054186582565308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,32,balanced,0.19732799132665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,32,balanced,0.4788159926732381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,32,balanced,0.2590879996617635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,32,balanced,0.6565386851628622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,32,1,power_law_1.2,0.3256959915161133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,32,1,power_law_1.01,0.50349440574646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,32,power_law_1.01,0.026931199431419372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,32,1,power_law_1.2,0.08315520286560059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,32,balanced,0.958725372950236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,32,1,power_law_1.2,0.350547194480896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,32,1,power_law_1.2,0.1390720009803772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,32,power_law_1.01,0.028755199909210206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,32,1,power_law_1.2,0.08661760091781616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,32,1,power_law_1.2,0.14043519496917725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,32,1,power_law_1.01,0.6342336177825928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,32,balanced,1.2600373427073162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,32,power_law_1.01,0.029190400242805482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,32,1,power_law_1.2,0.18302079439163207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,32,1,power_law_1.2,0.08866559863090515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,32,power_law_1.01,0.028441599011421202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,32,power_law_1.01,0.02768000066280365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,32,1,power_law_1.2,0.18889600038528442
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,32,1,power_law_1.2,0.09397119879722596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,32,power_law_1.01,0.028678399324417115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,32,1,power_law_1.2,0.1903167963027954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,32,1,power_law_1.2,0.10276479721069336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,32,1,power_law_1.2,0.1993407964706421
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,32,1,power_law_1.2,0.1092479944229126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,32,1,power_law_1.2,0.20464639663696288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,32,1,power_law_1.2,0.1219264030456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,32,balanced,0.6924586296081543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,32,1,power_law_1.2,0.13571200370788575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,32,balanced,0.9050933519999186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,32,1,power_law_1.2,0.1446463942527771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,32,1,power_law_1.2,0.42177281379699705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,32,1,power_law_1.2,0.17152639627456664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,32,power_law_1.01,0.03461120128631592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,32,1,power_law_1.2,0.4915328025817871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,32,power_law_1.01,0.034867200255393985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,32,1,power_law_1.2,0.16558719873428346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,32,power_law_1.01,0.0286655992269516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,32,1,power_law_1.01,0.8801728248596191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,32,power_law_1.01,0.03497599959373474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,32,power_law_1.01,0.031142398715019226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,32,1,power_law_1.2,0.19678080081939697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,32,power_law_1.01,0.02868480086326599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,32,1,power_law_1.2,0.5795199871063232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,32,power_law_1.01,0.0351936012506485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,32,power_law_1.01,0.03519999980926514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,32,power_law_1.01,0.028518399596214293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,32,1,power_law_1.2,0.2106559991836548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,32,power_law_1.01,0.03562879860401154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,32,power_law_1.01,0.03560320138931274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,32,1,power_law_1.01,1.16244478225708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,32,power_law_1.01,0.02874239981174469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,32,1,power_law_1.2,0.6714623928070068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,32,power_law_1.01,0.03697279989719391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,32,power_law_1.01,0.037964800000190736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,32,1,power_law_1.2,0.2540544033050537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,32,1,power_law_1.2,0.21521279811859131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,32,power_law_1.01,0.037574398517608645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,32,power_law_1.01,0.0383103996515274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,32,1,power_law_1.2,0.2297663927078247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,32,1,power_law_1.2,0.3968319892883301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,32,1,power_law_1.2,0.8748543739318848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,32,power_law_1.01,0.03889279961585999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,32,1,power_law_1.2,0.3121344089508057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,32,power_law_1.01,0.052697598934173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,32,1,power_law_1.2,0.4462592124938965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,32,power_law_1.01,0.05262719988822937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,32,1,power_law_1.2,0.3358720064163208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,32,1,power_law_1.2,1.1048319816589356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,32,power_law_1.01,0.05384960174560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,32,1,power_law_1.2,0.5098752021789551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,32,1,power_law_1.2,0.4114048004150391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,32,power_law_1.01,0.05601919889450073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,32,power_law_1.01,0.05800319910049438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,32,1,power_law_1.2,0.6406655788421631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,32,1,power_law_1.2,0.4565631866455078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,32,1,power_law_1.2,1.535142421722412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,32,power_law_1.01,0.06217600107192993
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,32,power_law_1.01,0.024038399755954742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,32,power_law_1.01,0.06741120219230652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,32,power_law_1.01,0.030713599920272828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,32,power_law_1.01,0.02372480034828186
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,32,1,power_law_1.2,0.8837375640869141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,32,power_law_1.01,0.033670398592948916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,32,power_law_1.01,0.07278079986572265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,32,power_law_1.01,0.043372800946235655
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,32,power_law_1.01,0.023660799860954283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,32,power_law_1.01,0.03444480001926422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,32,power_law_1.01,0.08306559920310974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,32,power_law_1.01,0.045542401075363156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,32,power_law_1.01,0.03739520013332367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,32,power_law_1.01,0.023955200612545014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,32,power_law_1.01,0.10561280250549317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,32,power_law_1.01,0.04628480076789856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,32,1,power_law_1.2,1.1874112129211425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,32,power_law_1.01,0.03768959939479828
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,32,power_law_1.01,0.02487040013074875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,32,power_law_1.01,0.049184000492095946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,32,power_law_1.01,0.12503679990768432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,32,power_law_1.01,0.04481280148029328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,32,power_law_1.2,0.0271232008934021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,32,power_law_1.01,0.052102398872375486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,32,power_law_1.01,0.17193599939346313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,32,power_law_1.2,0.027852800488471986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,32,power_law_1.01,0.058297598361968996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,32,power_law_1.01,0.21583359241485595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,32,power_law_1.2,0.02839680016040802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,32,power_law_1.01,0.0675711989402771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,32,power_law_1.01,0.08136320114135742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,32,power_law_1.2,0.028262400627136232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,32,power_law_1.01,0.3090431928634644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,32,1,power_law_1.2,0.5571263790130615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,32,power_law_1.01,0.09559040069580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,32,power_law_1.2,0.02813439965248108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,32,power_law_1.01,0.39301118850708006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,32,power_law_1.01,0.11853439807891845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,32,1,power_law_1.2,0.6629695892333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,32,power_law_1.01,0.14869120121002197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,32,1,power_law_1.2,1.9681215286254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,32,power_law_1.01,0.5729343891143799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,32,power_law_1.01,0.20937600135803222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,32,power_law_1.2,0.027353599667549133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,32,1,power_law_1.2,0.8751104354858399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,32,power_law_1.01,0.743500804901123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,32,power_law_1.01,0.25892479419708253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,32,power_law_1.2,0.02812800109386444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,32,power_law_1.01,0.027270400524139406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,32,power_law_1.01,0.04815999865531921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,32,power_law_1.2,0.028537601232528687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,32,power_law_1.01,0.3675904035568237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,32,power_law_1.01,0.028896000981330872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,32,power_law_1.01,0.04999679923057556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,32,1,power_law_1.2,1.0983615875244142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,32,power_law_1.2,0.027667200565338133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,32,power_law_1.01,1.0902400016784668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,32,power_law_1.01,0.052960002422332765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,32,power_law_1.01,0.4775424003601074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,32,power_law_1.2,0.02869119942188263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,32,power_law_1.01,0.05711359977722168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,32,power_law_1.2,0.0310591995716095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,32,power_law_1.01,0.061875200271606444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,32,power_law_1.01,0.6931007862091064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,32,power_law_1.2,0.03400959968566895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,32,1,power_law_1.2,1.5392704010009766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,32,power_law_1.2,0.030432000756263733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,32,power_law_1.01,0.07004160284996033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,32,power_law_1.01,1.4437631607055663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,32,power_law_1.2,0.03426559865474701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,32,power_law_1.2,0.03457919955253601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,32,power_law_1.01,0.08116480112075805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,32,power_law_1.2,0.03758080005645752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,32,power_law_1.2,0.03504000008106232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,32,power_law_1.01,0.9213567733764648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,32,power_law_1.2,0.03454720079898834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,32,power_law_1.01,0.09911680221557617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,32,power_law_1.2,0.03680639863014221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,32,power_law_1.2,0.03771519958972931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,32,power_law_1.2,0.03472639918327332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,32,power_law_1.01,0.1298367977142334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,32,power_law_1.2,0.036620798707008365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,32,power_law_1.2,0.04462080001831055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,32,power_law_1.2,0.03511680066585541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,32,power_law_1.01,0.1577407956123352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,32,power_law_1.2,0.04348160028457641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,32,power_law_1.2,0.047884801030159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,32,power_law_1.2,0.035097599029541016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,32,power_law_1.2,0.04459519982337952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,32,power_law_1.01,0.221343994140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,32,power_law_1.2,0.04952319860458374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,32,power_law_1.01,0.031411200761795044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,32,power_law_1.2,0.03606399893760681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,32,power_law_1.2,0.052857601642608644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,32,power_law_1.01,0.27779839038848875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,32,power_law_1.01,0.036345601081848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,32,power_law_1.2,0.03813120126724243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,32,power_law_1.2,0.05691519975662231
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,32,power_law_1.01,0.036800000071525577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,32,power_law_1.01,0.3993855953216553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,32,power_law_1.2,0.06168320178985596
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,32,power_law_1.01,0.037241598963737486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,32,power_law_1.2,0.06939520239830017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,32,power_law_1.01,0.5091392040252686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,32,power_law_1.01,0.042028799653053284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,32,power_law_1.2,0.08224639892578126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,32,power_law_1.2,0.023193599283695222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,32,power_law_1.01,0.040275201201438904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,32,power_law_1.2,0.09747200012207032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,32,power_law_1.2,0.022892799973487855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,32,power_law_1.01,0.7409855842590332
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,32,power_law_1.01,0.0367680013179779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,32,power_law_1.2,0.12931840419769286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,32,power_law_1.2,0.023219199478626253
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,32,power_law_1.01,0.037062400579452516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,32,power_law_1.2,0.1611199975013733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,32,power_law_1.2,0.02311680018901825
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,32,power_law_1.01,0.03752320110797882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,32,power_law_1.2,0.2224384069442749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,32,power_law_1.2,0.046137601137161255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,32,power_law_1.2,0.024300800263881685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,32,power_law_1.01,0.037401598691940305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,32,power_law_1.2,0.04883840084075928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,32,power_law_1.2,0.28279039859771726
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,32,power_law_1.2,0.027001601457595826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,32,power_law_1.01,0.041606399416923526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,32,power_law_1.2,0.0525056004524231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,32,power_law_1.2,0.038361600041389464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,32,1,power_law_1.2,1.9780223846435547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,32,power_law_1.2,0.028883200883865357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,32,balanced,0.03533866753180822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,32,power_law_1.2,0.05907840132713318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,32,balanced,0.03565866748491923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,32,power_law_1.2,0.0392767995595932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,32,balanced,0.03561066587766012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,32,power_law_1.2,0.030777600407600404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,32,balanced,0.03958933303753535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,32,power_law_1.2,0.06747519969940186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,32,balanced,0.056074668963750206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,32,power_law_1.2,0.05281280279159546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,32,power_law_1.2,0.035820800065994265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,32,power_law_1.2,0.08312960267066956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,32,power_law_1.2,0.0531711995601654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,32,power_law_1.2,0.09761919975280761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,32,power_law_1.2,0.05516160130500793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,32,power_law_1.2,0.12462079524993896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,32,power_law_1.2,0.056678402423858645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,32,power_law_1.2,0.15247360467910767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,32,power_law_1.2,0.05868160128593445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,32,power_law_1.01,0.9690943717956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,32,balanced,0.03378133227427801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,32,balanced,0.03436266630887985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,32,power_law_1.2,0.0625216007232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,32,balanced,0.03333866596221924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,32,power_law_1.2,0.21457281112670898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,32,balanced,0.035616000493367515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,32,balanced,0.04153066625197729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,32,power_law_1.2,0.06702719926834107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,32,power_law_1.2,0.07267199754714966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,32,power_law_1.2,0.08432639837265014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,32,balanced,0.059674665331840515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,32,balanced,0.06673599779605865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,32,power_law_1.2,0.10682239532470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,32,balanced,0.068271999557813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,32,balanced,0.08133333424727122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,32,power_law_1.2,0.03611519932746887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,32,balanced,0.08252800007661183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,32,power_law_1.2,0.1324031949043274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,32,balanced,0.09050666292508443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,32,balanced,0.09423466523488362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,32,power_law_1.2,0.036185601353645326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,32,balanced,0.09774399797121684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,32,balanced,0.10460266470909119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,32,balanced,0.1093280017375946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,32,power_law_1.2,0.040403199195861814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,32,balanced,0.11518399914105733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,32,balanced,0.12359467148780823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,32,balanced,0.1462559998035431
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,32,power_law_1.2,0.039680001139640805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,32,balanced,0.1665013333161672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,32,balanced,0.20354666312535605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,32,power_law_1.2,0.0359935998916626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,32,balanced,0.24078933397928873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,32,balanced,0.045552000403404236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,32,balanced,0.05420266588528951
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,32,power_law_1.2,0.036959999799728395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,32,balanced,0.33963199456532794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,32,balanced,0.05389333268006643
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,32,power_law_1.01,0.043987199664115906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,32,balanced,0.058330665032068886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,32,balanced,0.4203146696090698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,32,balanced,0.059877331058184304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,32,power_law_1.2,0.03695360124111176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,32,balanced,0.07349333167076111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,32,power_law_1.01,0.0500544011592865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,32,balanced,0.07875733574231465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,32,balanced,0.5969119866689047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,32,balanced,0.08339200417200725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,32,power_law_1.2,0.037452799081802365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,32,balanced,0.09120532870292664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,32,balanced,0.09708799918492635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,32,balanced,0.10498666763305664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,32,balanced,0.7708000342051188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,32,power_law_1.2,0.04067200124263763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,32,balanced,0.11725333333015442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,32,balanced,0.13779200116793314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,32,power_law_1.2,0.27394559383392336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,32,power_law_1.2,0.04550400078296661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,32,balanced,0.162063995997111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,32,balanced,1.125162680943807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,32,power_law_1.2,0.18437119722366332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,32,balanced,0.2092693249384562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,32,balanced,0.2545386751492818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,32,balanced,0.3563786745071411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,32,power_law_1.2,0.22586240768432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,32,balanced,1.4746987024943035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,32,balanced,0.44516801834106445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,32,balanced,0.0455626646677653
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,32,power_law_1.2,0.052537602186203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,32,balanced,0.04677866895993551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,32,balanced,0.045935998360315956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,32,balanced,0.6342613299687704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,32,balanced,0.04714133342107137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,32,balanced,0.056159997979799904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,32,power_law_1.2,0.05952640175819397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,32,balanced,0.0642080008983612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,32,balanced,0.06738666693369548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,32,balanced,0.0699893335501353
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,32,power_law_1.2,0.07625600099563598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,32,balanced,0.09583466251691182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,32,balanced,0.0993280013402303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,32,balanced,0.10296533505121867
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,32,power_law_1.2,0.09525120258331299
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,32,power_law_1.01,0.058310401439666745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,32,balanced,0.10730666915575664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,32,balanced,0.1125386655330658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,32,balanced,0.12106133500734965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,32,power_law_1.01,0.0719871997833252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,32,power_law_1.2,0.3895488023757935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,32,power_law_1.01,0.09057279825210571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,32,power_law_1.2,0.32556159496307374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,32,power_law_1.01,0.1288831949234009
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,32,power_law_1.01,0.16247040033340454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,32,1,power_law_1.2,0.6564095973968506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,32,power_law_1.01,0.23292160034179688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,32,balanced,0.8200480143229166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,32,power_law_1.01,0.3014400005340576
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,32,power_law_1.2,0.13992960453033448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,32,power_law_1.2,0.41235837936401365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,32,balanced,1.1925066312154133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,32,power_law_1.01,0.05301759839057922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,32,balanced,0.12784000237782797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,32,power_law_1.01,0.053414398431777955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,32,balanced,0.13617600003878275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,32,balanced,0.1518186628818512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,32,balanced,1.5655786196390789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,32,power_law_1.01,0.0532800018787384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,32,power_law_1.01,0.039052799344062805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,32,power_law_1.01,0.053452801704406736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,32,power_law_1.01,0.03933440148830414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,32,power_law_1.01,0.04961279928684235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,32,power_law_1.01,0.039724799990653994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,32,power_law_1.01,0.03935999870300293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,32,power_law_1.01,0.0421887993812561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,32,power_law_1.01,0.043110400438308716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,32,power_law_1.01,0.052179199457168576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,32,power_law_1.01,0.05423359870910645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,32,power_law_1.01,0.05790079832077026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,32,balanced,0.025226667523384094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,32,balanced,0.025231999655564625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,32,power_law_1.01,0.05719040036201477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,32,balanced,0.026522666215896606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,32,balanced,0.027263998985290527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,32,power_law_1.2,0.4046976089477539
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,32,balanced,0.029557332396507263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,32,power_law_1.01,0.0705344021320343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,32,balanced,0.03226666649182638
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,32,balanced,0.035418666899204254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,32,power_law_1.01,0.07668480277061462
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,32,balanced,0.03580799947182337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,32,balanced,0.08178666730721791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,32,power_law_1.01,0.08228480219841003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,32,balanced,0.0804319977760315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,32,balanced,0.06618133187294006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,32,balanced,0.055914665261904396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,32,power_law_1.01,0.09092479944229126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,32,balanced,0.04663466910521189
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,32,balanced,0.04769066472848257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,32,power_law_1.01,0.09882879853248597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,32,balanced,0.049770668148994446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,32,balanced,0.04977599779764811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,32,power_law_1.2,0.601580810546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,32,balanced,0.04762666424115499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,32,power_law_1.01,0.10579839944839478
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,32,balanced,0.0518453319867452
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,32,balanced,0.05601066847642263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,32,power_law_1.01,0.12336000204086303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,32,balanced,0.06623466809590657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,32,balanced,0.08196799953778584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,32,balanced,0.18090667327245077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,32,power_law_1.01,0.14387840032577515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,32,balanced,0.2111253341039022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,32,power_law_1.2,0.5025728225708008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,32,balanced,0.2967360019683838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,32,balanced,0.36419200897216797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,32,balanced,0.5150560140609741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,32,balanced,0.6616533199946085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,32,balanced,0.9595626990000407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,32,balanced,1.2534560362497966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,32,balanced,0.11748799681663513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,32,balanced,0.141567995150884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,32,balanced,1.8458773295084636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,32,balanced,0.19649600982666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,32,balanced,0.2555573383967082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,32,balanced,0.36129601796468097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,32,power_law_1.01,0.0574783980846405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,32,power_law_1.01,0.06567680239677429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,32,balanced,2.436896006266276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,32,power_law_1.01,0.0643455982208252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,32,power_law_1.01,0.055206400156021115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,32,power_law_1.01,0.08119040131568908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,32,power_law_1.01,0.05609599947929382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,32,power_law_1.01,0.08083840012550354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,32,power_law_1.01,0.05607039928436279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,32,power_law_1.01,0.08856319785118102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,32,power_law_1.01,0.09344639778137206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,32,power_law_1.01,0.054713600873947145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,32,power_law_1.01,0.09731839895248413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,32,power_law_1.01,0.05443840026855469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,32,power_law_1.01,0.1743232011795044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,32,power_law_1.01,0.10406399965286255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,32,power_law_1.01,0.06314240097999572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,32,power_law_1.01,0.22947840690612792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,32,power_law_1.01,0.1095039963722229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,32,power_law_1.01,0.06318079829216003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,32,power_law_1.01,0.11824640035629272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,32,power_law_1.01,0.06627200245857238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,32,balanced,0.4678186575571696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,32,power_law_1.01,0.1295232057571411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,32,power_law_1.01,0.09617279767990113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,32,power_law_1.01,0.15432319641113282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,32,power_law_1.01,0.09783679842948914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,32,power_law_1.01,0.17958400249481202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,32,power_law_1.01,0.23808000087738038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,32,power_law_1.01,0.2923648118972778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,32,power_law_1.01,0.40590720176696776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,32,power_law_1.01,0.5133056163787841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,32,power_law_1.01,0.2814079999923706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,32,power_law_1.01,0.3874560117721558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,32,power_law_1.01,0.7321728229522705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,32,power_law_1.01,0.10193920135498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,32,power_law_1.01,0.10558719635009765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,32,power_law_1.01,1.0218303680419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,32,power_law_1.01,0.10996479988098144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,32,power_law_1.01,0.12253439426422119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,32,power_law_1.01,1.389452838897705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,32,power_law_1.01,0.12949759960174562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,32,power_law_1.01,0.024806399643421174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,32,power_law_1.01,0.14128639698028564
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,32,power_law_1.01,0.025696000456809996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,32,power_law_1.2,0.7690048217773438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,32,power_law_1.01,0.15604480504989623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,32,power_law_1.01,0.02627840042114258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,32,power_law_1.01,0.20949759483337402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,32,power_law_1.01,1.8283647537231444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,32,power_law_1.01,0.026284798979759216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,32,power_law_1.01,0.25505919456481935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,32,power_law_1.2,0.05281919836997986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,32,power_law_1.01,0.027833598852157592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,32,power_law_1.2,0.051648002862930295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,32,power_law_1.01,0.03049600124359131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,32,power_law_1.2,0.04886400103569031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,32,power_law_1.01,0.033036801218986514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,32,power_law_1.2,0.049004799127578734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,32,power_law_1.01,0.03389439880847931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,32,power_law_1.2,0.05124480128288269
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,32,power_law_1.01,0.06826239824295044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,32,power_law_1.2,0.05626879930496216
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,32,power_law_1.01,0.0706496000289917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,32,power_law_1.2,0.06425600051879883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,32,power_law_1.01,0.058316802978515624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,32,power_law_1.2,0.06581760048866273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,32,power_law_1.01,0.04916479885578155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,32,power_law_1.2,0.0800383985042572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,32,power_law_1.01,0.4873983860015869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,32,power_law_1.2,0.08151040077209473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,32,power_law_1.01,0.34622080326080323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,32,power_law_1.01,0.7048895835876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,32,power_law_1.01,0.4340735912322998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,32,power_law_1.01,0.6154623985290527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,32,power_law_1.01,0.9384960174560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,32,power_law_1.2,0.5236032009124756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,32,power_law_1.01,0.8231167793273926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,32,power_law_1.01,1.3447423934936524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,32,power_law_1.01,1.1940095901489258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,32,power_law_1.01,0.043007999658584595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,32,power_law_1.2,0.08778240084648133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,32,power_law_1.01,0.044537600874900815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,32,power_law_1.2,0.09245439767837524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,32,power_law_1.01,1.8293952941894531
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,32,power_law_1.01,0.04572800099849701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,32,power_law_1.2,0.09761919975280761
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,32,power_law_1.01,0.04860160052776337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,32,power_law_1.01,1.5368255615234374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,32,power_law_1.2,0.03899520039558411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,32,power_law_1.2,0.10487040281295776
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,32,power_law_1.01,0.04805119931697845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,32,power_law_1.2,0.03945600092411041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,32,power_law_1.2,0.11111680269241334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,32,power_law_1.01,0.056992000341415404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,32,power_law_1.2,0.03956480026245117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,32,power_law_1.2,0.11767040491104126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,32,power_law_1.01,0.06624640226364135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,32,power_law_1.2,0.036620798707008365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,32,power_law_1.2,0.13070080280303956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,32,power_law_1.01,2.3448448181152344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,32,power_law_1.01,0.08415359854698182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,32,power_law_1.2,0.04106239974498749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,32,power_law_1.2,0.16277120113372803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,32,power_law_1.2,0.04447999894618988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,32,power_law_1.01,0.10097919702529908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,32,power_law_1.2,0.19465600252151488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,32,power_law_1.2,0.052928000688552856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,32,power_law_1.01,0.1319424033164978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,32,power_law_1.2,0.24995200634002684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,32,power_law_1.2,0.054022401571273804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,32,power_law_1.01,0.164300799369812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,32,power_law_1.2,0.05660799741744995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,32,power_law_1.2,0.3175807952880859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,32,power_law_1.2,0.059059202671051025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,32,power_law_1.01,0.24005119800567626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,32,power_law_1.01,3.2423423767089843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,32,power_law_1.2,0.06924160122871399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,32,power_law_1.2,0.4271423816680908
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,32,power_law_1.01,0.31597440242767333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,32,power_law_1.2,0.054662400484085084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,32,power_law_1.2,0.07603840231895446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,32,power_law_1.2,0.5350272178649902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,32,power_law_1.01,0.4343167781829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,32,power_law_1.01,0.6048831939697266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,32,power_law_1.2,0.7443840026855468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,32,power_law_1.2,0.08268799781799316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,32,power_law_1.2,0.09028480052947999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,32,power_law_1.2,0.09973120093345642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,32,power_law_1.2,0.10622080564498901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,32,power_law_1.2,0.12625919580459594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,32,power_law_1.2,0.15063040256500243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,32,power_law_1.2,0.17850879430770875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,32,power_law_1.2,0.24331519603729249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,32,power_law_1.2,0.05600000023841858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,32,power_law_1.2,0.2975231885910034
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,32,power_law_1.2,0.02515200078487396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,32,power_law_1.2,0.055667197704315184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,32,power_law_1.2,0.17697279453277587
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,32,power_law_1.2,0.025740799307823182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,32,power_law_1.2,0.41127681732177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,32,power_law_1.2,0.05548800230026245
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,32,power_law_1.2,0.026022401452064515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,32,power_law_1.2,0.05565440058708191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,32,power_law_1.2,0.5102911949157715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,32,power_law_1.2,0.06589440107345582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,32,power_law_1.2,0.026105600595474242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,32,power_law_1.2,0.06425600051879883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,32,power_law_1.2,0.02717440128326416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,32,power_law_1.2,0.06955519914627076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,32,power_law_1.2,1.1497599601745605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,32,power_law_1.2,0.030873599648475646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,32,power_law_1.2,0.09618560075759888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,32,power_law_1.2,0.03372159898281098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,32,power_law_1.2,0.09743360280990601
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,32,power_law_1.2,0.03433600068092346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,32,power_law_1.2,0.10136959552764893
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,32,power_law_1.2,0.07285119891166687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,32,power_law_1.2,0.10654079914093018
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,32,power_law_1.2,0.07144320011138916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,32,power_law_1.2,0.11164799928665162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,32,power_law_1.2,0.05804160237312317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,32,power_law_1.2,0.1220736026763916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,32,power_law_1.2,0.05088000297546387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,32,power_law_1.2,0.12968319654464722
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,32,power_law_1.2,0.04323840141296387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,32,power_law_1.2,0.14043519496917725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,32,power_law_1.2,0.04310399889945984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,32,power_law_1.2,0.1608896017074585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,32,power_law_1.2,0.047705599665641786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,32,power_law_1.2,0.2182784080505371
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,32,power_law_1.2,0.04897280037403107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,32,power_law_1.2,0.26670079231262206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,32,power_law_1.2,0.04903680086135864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,32,power_law_1.2,0.770579195022583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,32,power_law_1.2,0.3686975955963135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,32,power_law_1.2,0.5080383777618408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,32,power_law_1.2,0.7340415954589844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,32,power_law_1.2,0.6566080093383789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,32,power_law_1.2,0.060288000106811526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,32,power_law_1.2,0.066048002243042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,32,power_law_1.2,0.08643199801445008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,32,power_law_1.2,0.9155008316040039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,32,power_law_1.2,0.10907520055770874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,32,power_law_1.2,0.14565759897232056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,32,power_law_1.2,0.17727999687194823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,32,power_law_1.2,1.2695551872253419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,32,power_law_1.2,1.0284159660339356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,32,power_law_1.2,0.25064959526062014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,32,power_law_1.2,1.0089152336120606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,32,power_law_1.2,0.9840767860412598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,32,1,balanced,0.029178666571776073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,32,1,balanced,0.03346133232116699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,32,1,balanced,0.041509332756201424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,32,1,balanced,0.06425599753856659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,32,1,balanced,0.10284266869227092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,32,1,balanced,0.10508267084757487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,32,1,balanced,0.10459199547767639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,32,1,balanced,0.10422933101654053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,32,1,balanced,0.08673600355784099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,32,1,balanced,0.08749866485595703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,32,1,balanced,0.10319466392199199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,32,1,balanced,0.10320533315340678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,32,1,balanced,0.10517332951227824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,32,1,balanced,0.10995200276374817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,32,1,balanced,0.11143466830253601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,32,1,balanced,0.11716799934705098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,32,1,balanced,0.1237440009911855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,32,1,balanced,0.1551520029703776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,32,1,balanced,0.1674506664276123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,32,1,balanced,0.19097065925598145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,32,1,balanced,0.2140586574872335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,32,1,balanced,0.2964266737302144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,32,1,balanced,0.3465493520100911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,32,1,balanced,0.5081813335418701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,32,1,balanced,0.640064001083374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,32,power_law_1.2,1.7645439147949218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,32,1,balanced,0.9419093132019043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,32,1,balanced,1.2446346282958984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,32,1,balanced,0.025386666258176167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,32,1,balanced,0.027210667729377747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,32,1,balanced,0.03154666721820831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,32,1,balanced,0.04162133236726125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,32,1,balanced,0.05788266658782959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,32,1,balanced,0.060309335589408875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,32,1,balanced,0.07226666808128357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,32,1,balanced,0.07283733288447063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,32,1,balanced,0.07251733541488647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,32,1,balanced,0.07222400108973186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,32,1,balanced,0.08453333377838135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,32,1,balanced,0.0885706643263499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,32,1,balanced,0.09082133571306865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,32,1,balanced,0.09471467137336731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,32,1,balanced,0.10212799906730652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,32,1,balanced,0.10636267066001892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,32,1,balanced,0.11526399850845337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,32,1,balanced,0.1476479967435201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,32,1,balanced,0.16132266322771707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,32,1,balanced,0.1890559991200765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,32,1,balanced,0.2169119914372762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,32,1,balanced,0.31011199951171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,32,1,balanced,0.3642186721165975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,32,1,balanced,0.5252960125605265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,32,1,balanced,0.6720639864603678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,32,1,balanced,0.9897706508636475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,32,1,balanced,1.3074346383412678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,32,1,power_law_1.01,0.027564799785614012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,32,1,power_law_1.01,0.03155199885368347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,32,1,power_law_1.01,0.04183039963245392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,32,1,power_law_1.01,0.05390080213546753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,32,power_law_1.2,0.771731185913086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,32,1,power_law_1.01,0.06551039814949036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,32,1,power_law_1.01,0.09102079868316651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,32,1,power_law_1.01,0.09502080082893372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,32,1,power_law_1.01,0.09785599708557129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,32,1,power_law_1.01,0.0819711983203888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,32,1,power_law_1.01,0.08388479948043823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,32,1,power_law_1.01,0.09953280091285706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,32,1,power_law_1.01,0.10038399696350098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,32,1,power_law_1.01,0.10273280143737792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,32,1,power_law_1.01,0.10802559852600098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,32,1,power_law_1.01,0.11260160207748413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,32,1,power_law_1.01,0.1168511986732483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,32,1,power_law_1.01,0.12696319818496704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,32,1,power_law_1.01,0.16138880252838134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,32,1,power_law_1.01,0.17827839851379396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,32,1,power_law_1.01,0.21154561042785644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,32,1,power_law_1.01,0.24208641052246094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,32,1,power_law_1.01,0.30872321128845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,32,power_law_1.2,0.33041279315948485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,32,power_law_1.2,0.25428481101989747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,32,1,power_law_1.01,0.4018752098083496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,32,1,power_law_1.01,0.5433919906616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,32,1,power_law_1.01,0.7072127819061279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,32,1,power_law_1.01,0.9884096145629883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,32,1,power_law_1.01,1.2978816032409668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,32,1,power_law_1.01,0.02369920015335083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,32,1,power_law_1.01,0.02560639977455139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,32,1,power_law_1.01,0.029254400730133058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,32,1,power_law_1.01,0.03684479892253876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,32,1,power_law_1.01,0.043993601202964784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,32,1,power_law_1.01,0.05489919781684875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,32,1,power_law_1.01,0.06762239933013917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,32,1,power_law_1.01,0.07092480063438415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,32,1,power_law_1.01,0.06921600103378296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,32,1,power_law_1.01,0.0692736029624939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,32,1,power_law_1.01,0.08324480056762695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,32,1,power_law_1.01,0.08595200181007386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,32,1,power_law_1.01,0.09084799885749817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,32,1,power_law_1.01,0.09615359902381897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,32,1,power_law_1.01,0.10193920135498047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,32,1,power_law_1.01,0.10823680162429809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,32,1,power_law_1.01,0.11847679615020752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,32,1,power_law_1.01,0.15231360197067262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,32,1,power_law_1.01,0.1741055965423584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,32,1,power_law_1.01,0.21008639335632323
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,32,1,power_law_1.01,0.24773120880126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,32,1,power_law_1.01,0.3219583988189697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,32,1,power_law_1.01,0.4096127986907959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,32,power_law_1.2,0.474124813079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,32,1,power_law_1.01,0.561740779876709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,32,1,power_law_1.01,0.7241856098175049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,32,1,power_law_1.01,1.03504638671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,32,1,power_law_1.01,1.3582015991210938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,32,1,power_law_1.2,0.02752000093460083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,32,1,power_law_1.2,0.031615999341011045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,32,1,power_law_1.2,0.04139519929885864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,32,1,power_law_1.2,0.050246399641036985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,32,1,power_law_1.2,0.06165120005607605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,32,1,power_law_1.2,0.09040639996528625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,32,1,power_law_1.2,0.09573119878768921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,32,1,power_law_1.2,0.0985472023487091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,32,1,power_law_1.2,0.08395519852638245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,32,1,power_law_1.2,0.08485119938850402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,32,1,power_law_1.2,0.10115840435028076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,32,1,power_law_1.2,0.10081280469894409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,32,1,power_law_1.2,0.10451840162277222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,32,1,power_law_1.2,0.10981760025024415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,32,1,power_law_1.2,0.1146239995956421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,32,1,power_law_1.2,0.1192512035369873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,32,1,power_law_1.2,0.12947839498519897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,32,1,power_law_1.2,0.16314879655838013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,32,1,power_law_1.2,0.18198399543762206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,32,power_law_1.2,2.506412887573242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,32,1,power_law_1.2,0.21374080181121827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,32,1,power_law_1.2,0.24661760330200194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,32,1,power_law_1.2,0.3102911949157715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,32,1,power_law_1.2,0.40555520057678224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,32,1,power_law_1.2,0.5470975875854492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,32,power_law_1.2,3.264691162109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,32,1,power_law_1.2,0.7115071773529053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,32,1,power_law_1.2,0.02351360023021698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,32,1,power_law_1.2,0.02588160037994385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,32,1,power_law_1.2,0.02961280047893524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,32,1,power_law_1.2,0.037964800000190736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,32,1,power_law_1.2,0.042156800627708435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,32,1,power_law_1.2,0.055206400156021115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,32,1,power_law_1.2,0.06869760155677795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,32,1,power_law_1.2,0.07033600211143494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,32,1,power_law_1.2,0.0686847984790802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,32,1,power_law_1.2,0.07121279835700989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,32,1,power_law_1.2,0.08201599717140198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,32,1,power_law_1.2,0.08593279719352723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,32,1,power_law_1.2,0.09318400025367737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,32,1,power_law_1.2,0.097952002286911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,32,1,power_law_1.2,0.10269440412521362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,32,1,power_law_1.2,0.10920959711074829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,32,1,power_law_1.2,0.12015999555587768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,32,1,power_law_1.2,0.15628160238265992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,32,1,power_law_1.2,0.17579519748687744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,32,1,power_law_1.2,0.21276159286499025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,32,1,power_law_1.2,0.24942080974578856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,32,power_law_1.2,0.6288832187652588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,32,balanced,0.04799999793370565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,32,balanced,0.04826133449872335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,32,balanced,0.047584002216657005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,32,balanced,0.06398933132489522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,32,1,power_law_1.2,0.3224319934844971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,32,balanced,0.09506133198738098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,32,balanced,0.11751466989517212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,32,balanced,0.12115200360616048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,32,balanced,0.12170666456222534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,32,balanced,0.13849066694577536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,32,balanced,0.13902933398882547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,32,balanced,0.15335466464360556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,32,balanced,0.1602186659971873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,32,balanced,0.16274666786193848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,32,balanced,0.1722453236579895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,32,balanced,0.18039999405543009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,32,balanced,0.18756266434987387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,32,balanced,0.20253332455952963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,32,balanced,0.2601173321406047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,32,balanced,0.29100267092386883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,32,balanced,0.3508373498916626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,32,balanced,0.41260798772176105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,32,1,power_law_1.2,0.4117440223693848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,32,balanced,0.5879093408584595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,32,balanced,0.7091786861419678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,32,balanced,1.0062026977539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,32,balanced,1.29639999071757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,32,balanced,1.8403040568033855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,32,balanced,2.446869373321533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,32,balanced,0.03945599993069967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,32,balanced,0.03774400055408478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,32,balanced,0.03952533255020777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,32,balanced,0.051818668842315674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,32,balanced,0.07047466437021892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,32,balanced,0.07870933413505554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,32,balanced,0.09856533010800679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,32,balanced,0.09777067104975383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,32,balanced,0.10461866855621338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,32,balanced,0.10763200124104817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,32,balanced,0.1256586710611979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,32,balanced,0.13805333773295084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,32,balanced,0.1418400009473165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,32,power_law_1.2,0.9937472343444824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,32,balanced,0.15363199512163797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,32,balanced,0.16394133369127908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,32,balanced,0.05806933343410492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,32,balanced,0.17292799552281699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,32,balanced,0.0595360000928243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,32,balanced,0.19521600008010864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,32,balanced,0.05858666698137919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,32,balanced,0.07260799904664357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,32,balanced,0.24180267254511514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,32,balanced,0.08680533369382222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,32,balanced,0.2739413380622864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,32,balanced,0.10518933335940044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,32,balanced,0.1106666624546051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,32,balanced,0.34277331829071045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,32,balanced,0.10795733332633972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,32,balanced,0.16457066933314005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,32,balanced,0.16827734311421713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,32,balanced,0.1763733426729838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,32,balanced,0.18141865730285645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,32,balanced,0.18939199050267538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,32,balanced,0.20293867588043213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,32,balanced,0.21848533550898233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,32,balanced,0.23219199975331625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,32,balanced,0.25780800978342694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,32,balanced,0.31015467643737793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,32,balanced,0.36274667580922443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,32,balanced,0.516543984413147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,32,balanced,0.6200906833012899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,32,1,power_law_1.2,0.5674431800842286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,32,balanced,0.4103200038274129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,32,balanced,0.8801706631978353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,32,balanced,0.5759520133336385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,32,balanced,1.1355520089467366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,32,balanced,0.7123306592305502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,32,balanced,1.0113813082377117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,32,balanced,1.5989866256713867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,32,balanced,1.3052639961242676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,32,balanced,2.106837272644043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,32,balanced,1.8733545939127605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,32,balanced,3.0813865661621094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,32,balanced,2.468992074330648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,32,balanced,4.048896153767903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,32,power_law_1.01,0.10379519462585449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,32,power_law_1.01,0.10367360115051269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,32,power_law_1.01,0.10368640422821045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,32,power_law_1.01,0.09285119771957398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,32,power_law_1.01,0.09540479779243469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,32,power_law_1.01,0.11277439594268798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,32,power_law_1.01,0.11251200437545776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,32,power_law_1.01,0.11364480257034301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,32,power_law_1.01,0.13496960401535035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,32,power_law_1.01,0.13479039669036866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,32,power_law_1.01,0.150655996799469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,32,power_law_1.01,0.15552639961242676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,32,power_law_1.01,0.1583295941352844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,32,power_law_1.01,0.1710592031478882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,32,power_law_1.01,0.18504960536956788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,32,power_law_1.01,0.20158720016479492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,32,power_law_1.01,0.2424704074859619
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,32,balanced,0.02924266705910365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,32,balanced,0.02916266769170761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,32,power_law_1.01,0.28395519256591795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,32,balanced,0.02917333443959554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,32,balanced,0.031285333136717476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,32,balanced,0.04158399999141693
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,32,balanced,0.05349333087603251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,32,power_law_1.01,0.32974720001220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,32,balanced,0.060415998101234436
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,32,balanced,0.08014399806658427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,32,balanced,0.07961600025494893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,32,power_law_1.01,0.43057918548583984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,32,balanced,0.1946559945742289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,32,power_law_1.2,1.4523136138916015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,32,balanced,0.1846239964167277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,32,balanced,0.16581333676973978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,32,power_law_1.01,0.5183872222900391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,32,1,power_law_1.2,0.732147216796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,32,power_law_1.01,0.7171840190887451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,32,power_law_1.01,0.8875328063964844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,32,power_law_1.01,1.2548864364624024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,32,balanced,0.1564479966958364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,32,balanced,0.09470400214195251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,32,balanced,0.09649599591890971
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,32,balanced,0.082096000512441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,32,balanced,0.08242666721343994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,32,balanced,0.08768000205357869
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,32,balanced,0.0971999963124593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,32,balanced,0.11224533120791118
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,32,balanced,0.1313973367214203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,32,balanced,0.17674134174982706
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,32,balanced,0.21596266825993857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,32,balanced,0.3046666582425435
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,32,balanced,0.3949600060780843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,32,balanced,0.5517866611480713
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,32,balanced,0.7464533646901449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,32,power_law_1.01,1.6685440063476562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,32,power_law_1.01,0.07214080095291138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,32,1,power_law_1.2,1.0427071571350097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,32,power_law_1.01,0.07179520130157471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,32,power_law_1.01,0.07185279726982116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,32,power_law_1.01,0.07300480008125305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,32,power_law_1.01,2.4474048614501953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,32,power_law_1.01,0.06981760263442993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,32,power_law_1.01,0.07463679909706115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,32,power_law_1.01,0.0942911982536316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,32,power_law_1.01,0.09683200120925903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,32,power_law_1.01,0.10072319507598877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,32,power_law_1.01,3.3653377532958983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,32,power_law_1.01,0.1034432053565979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,32,power_law_1.01,0.12419840097427368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,32,1,power_law_1.2,0.9935359954833984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,32,power_law_1.01,0.08627840280532836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,32,power_law_1.01,0.13598719835281373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,32,power_law_1.01,0.08736000061035157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,32,power_law_1.01,0.14006400108337402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,32,power_law_1.01,0.08844159841537476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,32,power_law_1.01,0.15033600330352784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,32,power_law_1.01,0.08672000169754028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,32,power_law_1.01,0.1734976053237915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,32,power_law_1.01,0.08379520177841186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,32,power_law_1.01,0.17959680557250976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,32,power_law_1.01,0.10243200063705445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,32,power_law_1.01,0.22069120407104492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,32,power_law_1.01,0.10643839836120605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,32,power_law_1.01,0.2568703889846802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,32,power_law_1.01,0.10493439435958862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,32,power_law_1.01,0.2901760101318359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,32,power_law_1.01,0.1614527940750122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,32,power_law_1.01,0.3773952007293701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,32,power_law_1.01,0.1641535997390747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,32,power_law_1.01,0.4799168109893799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,32,power_law_1.01,0.1738752007484436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,32,power_law_1.01,0.18062080144882203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,32,power_law_1.01,0.6350719928741455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,32,power_law_1.01,0.1850816011428833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,32,power_law_1.01,0.1988927960395813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,32,power_law_1.01,0.8102656364440918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,32,power_law_1.01,0.22801918983459474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,32,power_law_1.01,0.251475191116333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,32,power_law_1.01,1.1644415855407715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,32,power_law_1.01,0.3018239974975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,32,power_law_1.01,0.36970880031585696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,32,power_law_1.01,1.5130687713623048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,32,power_law_1.01,0.45862398147583006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,32,power_law_1.01,0.6283711910247802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,32,power_law_1.01,2.161759948730469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,32,power_law_1.01,0.7910079956054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,32,power_law_1.01,1.0891072273254394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,32,power_law_1.2,1.532857608795166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,32,power_law_1.01,2.8511423110961913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,32,power_law_1.2,1.518124771118164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,32,power_law_1.01,1.4244864463806153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,32,power_law_1.2,0.10270719528198242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,32,power_law_1.2,0.09655680060386658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,32,power_law_1.2,0.10310399532318115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,32,power_law_1.01,2.157139205932617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,32,power_law_1.01,2.8290431976318358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,32,power_law_1.2,0.08974720239639282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,32,power_law_1.2,0.0925055980682373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,32,power_law_1.2,0.1124735951423645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,32,power_law_1.01,4.057740783691406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,32,power_law_1.2,0.11455999612808228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,32,power_law_1.2,0.12184959650039673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,32,power_law_1.2,0.13342080116271973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,32,power_law_1.2,0.1349951982498169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,32,power_law_1.2,0.1473855972290039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,32,power_law_1.2,0.15545599460601806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,32,power_law_1.2,0.1637503981590271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,32,power_law_1.01,0.046028798818588255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,32,power_law_1.2,0.17834240198135376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,32,power_law_1.01,5.395660781860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,32,power_law_1.2,0.18791040182113647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,32,power_law_1.01,0.046387198567390445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,32,power_law_1.2,0.07297919988632202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,32,power_law_1.01,0.04741120040416717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,32,power_law_1.2,0.20979840755462648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,32,power_law_1.2,0.07245439887046815
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,32,power_law_1.01,0.041843199729919435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,32,power_law_1.2,0.2431999921798706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,32,power_law_1.2,0.07287039756774902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,32,power_law_1.01,0.04192639887332916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,32,power_law_1.2,0.069760000705719
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,32,power_law_1.01,0.05082880258560181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,32,power_law_1.2,0.06537600159645081
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,32,power_law_1.01,0.056428802013397214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,32,power_law_1.2,0.07597439885139465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,32,power_law_1.01,0.07004160284996033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,32,power_law_1.2,0.09386879801750184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,32,power_law_1.2,0.09644160270690919
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,32,power_law_1.01,0.07046399712562561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,32,power_law_1.2,0.10032639503479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,32,power_law_1.01,0.17468160390853882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,32,power_law_1.2,0.10393600463867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,32,power_law_1.01,0.15679999589920043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,32,power_law_1.2,0.12524800300598143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,32,power_law_1.01,0.14729599952697753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,32,power_law_1.2,0.30426878929138185
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,32,power_law_1.01,0.12421760559082032
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,32,power_law_1.01,0.08338559865951538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,32,power_law_1.2,0.33994240760803224
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,32,power_law_1.01,0.08289920091629029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,32,power_law_1.2,0.44992637634277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,32,power_law_1.01,0.0743615984916687
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,32,power_law_1.01,0.07948160171508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,32,power_law_1.01,0.0870527982711792
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,32,power_law_1.01,0.10127999782562255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,32,power_law_1.01,0.12675199508666993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,32,power_law_1.2,0.13422080278396606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,32,power_law_1.01,0.1513216018676758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,32,power_law_1.2,0.1394368052482605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,32,power_law_1.01,0.20700159072875976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,32,power_law_1.2,0.15561599731445314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,32,power_law_1.01,0.26396799087524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,32,power_law_1.2,0.17100800275802613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,32,power_law_1.2,0.19034240245819092
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,32,power_law_1.01,0.3798975944519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,32,power_law_1.2,0.21965439319610597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,32,power_law_1.01,0.5240128040313721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,32,power_law_1.2,0.2578687906265259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,32,power_law_1.2,0.3002624034881592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,32,power_law_1.01,0.7149824142456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,32,power_law_1.2,0.3987519979476929
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,32,power_law_1.01,0.9929856300354004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,32,power_law_1.2,0.545305585861206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,32,power_law_1.2,0.08556159734725952
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,32,power_law_1.2,0.34279680252075195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,32,power_law_1.2,0.08714240193367004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,32,power_law_1.2,0.8025728225708008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,32,power_law_1.2,0.08819839954376221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,32,power_law_1.2,0.08464000225067139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,32,power_law_1.2,0.08071680068969726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,32,power_law_1.2,0.10069760084152221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,32,power_law_1.2,0.48966398239135744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,32,power_law_1.2,1.9053312301635743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,32,power_law_1.2,0.10579839944839478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,32,1,balanced,0.035386666655540466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,32,1,balanced,0.04400533437728882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,32,power_law_1.2,0.10885119438171387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,32,1,balanced,0.055674667159716286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,32,1,balanced,0.0930560032526652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,32,1,balanced,0.15470932920773825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,32,power_law_1.2,0.1578112006187439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,32,power_law_1.2,0.6879551887512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,32,1,balanced,0.20558400948842367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,32,1,balanced,0.184879998366038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,32,power_law_1.2,0.1661247968673706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,32,1,balanced,0.18452799320220947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,32,1,balanced,0.1641813317934672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,32,power_law_1.2,0.17329280376434325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,32,power_law_1.2,0.8416831970214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,32,power_law_1.2,0.18110719919204712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,32,power_law_1.2,0.18981120586395264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,32,power_law_1.2,0.2162623882293701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,32,power_law_1.2,0.9985600471496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,32,power_law_1.2,0.24124159812927246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,32,power_law_1.2,0.26839680671691896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,32,power_law_1.2,0.29253759384155276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,32,power_law_1.2,1.3630144119262695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,32,power_law_1.2,0.39202558994293213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,32,1,balanced,0.16448533535003662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,32,1,balanced,0.1835306684176127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,32,1,balanced,0.18415466944376627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,32,1,balanced,0.18690667549769083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,32,power_law_1.2,0.5022143840789794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,32,1,balanced,0.19268266359965006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,32,1,balanced,0.19605867067972818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,32,1,balanced,0.19986667235692343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,32,1,balanced,0.21151467164357504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,32,1,balanced,0.3134933312733968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,32,1,balanced,0.3295573393503825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,32,1,balanced,0.3635093371073405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,32,1,power_law_1.2,1.3692095756530762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,32,power_law_1.2,0.04778240025043488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,32,1,balanced,0.029498666524887085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,32,power_law_1.2,1.222067165374756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,32,1,balanced,0.03334933271010717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,32,1,balanced,0.03941866755485535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,32,power_law_1.2,0.04759039878845215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,32,1,balanced,0.061861331264177956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,32,1,balanced,0.09325866897900899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,32,1,balanced,0.10935466488202412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,32,power_law_1.2,0.046623998880386354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,32,1,balanced,0.1346506675084432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,32,1,balanced,0.13807466626167297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,32,1,balanced,0.1448906660079956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,32,power_law_1.2,0.04270080029964447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,32,power_law_1.2,0.038227200508117676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,32,power_law_1.2,0.666764783859253
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,32,power_law_1.2,0.05060480237007141
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,32,power_law_1.2,0.05737599730491638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,32,power_law_1.2,0.8449855804443359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,32,power_law_1.2,0.07336320281028748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,32,1,balanced,0.39688531557718915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,32,power_law_1.2,0.07124480009078979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,32,1,balanced,0.5861599842707316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,32,1,balanced,0.6560800075531006
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,32,power_law_1.2,0.1553984045982361
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,32,power_law_1.2,0.1335935950279236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,32,1,balanced,0.9268799622853597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,32,power_law_1.2,1.2897472381591797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,32,1,balanced,0.14415466785430908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,32,1,balanced,0.15225600202878317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,32,power_law_1.2,0.1354367971420288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,32,1,balanced,0.15436800320943198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,32,1,balanced,0.1588320036729177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,32,1,balanced,1.196346680323283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,32,1,balanced,0.16222400466601053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,32,power_law_1.2,0.12856320142745972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,32,1,balanced,0.16898133357365927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,32,1,balanced,0.17377066612243652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,32,1,balanced,0.18343466520309448
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,32,power_law_1.2,0.08492159843444824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,32,1,balanced,1.6099200248718262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,32,1,balanced,0.2625173330307007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,32,1,balanced,0.27959465980529785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,32,power_law_1.2,0.08590720295906067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,32,power_law_1.2,1.6247743606567382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,32,1,balanced,0.31835200389226276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,32,1,balanced,0.35497601826985675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,32,1,balanced,2.1785759925842285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,32,1,balanced,0.5295360088348389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,32,1,power_law_1.01,0.03332479894161224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,32,power_law_1.2,1.5993535995483399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,32,1,power_law_1.01,0.04176000058650971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,32,power_law_1.2,2.3422719955444338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,32,1,power_law_1.01,0.05601919889450073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,32,1,power_law_1.01,0.08439040184020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,32,power_law_1.2,2.0307392120361327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,32,1,power_law_1.01,0.10409599542617798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,32,1,power_law_1.01,0.15403519868850707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,32,1,power_law_1.01,0.15565439462661743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,32,power_law_1.2,0.07420799732208253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,32,1,power_law_1.01,0.16670080423355102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,32,power_law_1.2,0.08127359747886657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,32,1,power_law_1.01,0.14420479536056519
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,32,power_law_1.2,0.09148799777030944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,32,1,power_law_1.01,0.14903680086135865
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,32,power_law_1.2,0.10517760515213012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,32,1,power_law_1.01,0.1744447946548462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,32,power_law_1.2,3.2661888122558596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,32,1,power_law_1.01,0.17704960107803344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,32,power_law_1.2,2.291833686828613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,32,1,power_law_1.01,0.17740800380706787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,32,1,power_law_1.01,0.18498560190200805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,32,1,power_law_1.01,0.19177600145339965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,32,1,power_law_1.01,0.20074241161346434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,32,1,power_law_1.01,0.2149888038635254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,32,1,balanced,0.6049066781997681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,32,1,power_law_1.01,0.3206720113754272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,32,power_law_1.2,4.530707168579101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,32,1,balanced,0.8572426637013754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,32,1,power_law_1.01,0.3457855939865112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,32,power_law_1.2,0.13073279857635497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,32,1,balanced,1.1058453718821208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,32,1,power_law_1.01,0.39467520713806153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,32,power_law_1.2,0.16330879926681519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,32,1,power_law_1.01,0.4446591854095459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,32,1,balanced,1.5160959561665852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,32,power_law_1.2,0.2167423963546753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,32,1,power_law_1.01,0.5810431957244873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,32,power_law_1.2,0.3019776105880737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,32,1,balanced,2.025541305541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,32,1,power_law_1.01,0.6921472072601318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,32,1,power_law_1.01,0.028384000062942505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,32,1,power_law_1.01,0.031513598561286923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,32,power_law_1.2,0.42197117805480955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,32,1,power_law_1.01,0.9348544120788574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,32,1,power_law_1.01,0.043833601474761966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,32,1,power_law_1.01,0.05782399773597717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,32,1,power_law_1.01,1.1893312454223632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,32,1,power_law_1.01,1.6781248092651366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,32,1,power_law_1.01,2.192127990722656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,32,1,power_law_1.2,0.03363839983940124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,32,1,power_law_1.01,0.06947199702262878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,32,power_law_1.2,3.2286655426025392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,32,1,power_law_1.2,0.04345600008964538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,32,1,power_law_1.01,0.09185919761657715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,32,1,power_law_1.01,0.11791360378265381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,32,1,power_law_1.2,0.05544959902763367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,32,1,power_law_1.2,0.029446399211883544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,32,1,power_law_1.01,0.12437119483947753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,32,1,power_law_1.2,0.07508479952812194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,32,1,power_law_1.2,0.03596799969673157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,32,1,power_law_1.01,0.1220352053642273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,32,1,power_law_1.2,0.09872639775276185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,32,1,power_law_1.01,0.13474559783935547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,32,power_law_1.2,6.4888771057128904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,32,1,power_law_1.2,0.1537279963493347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,32,balanced,0.05048533280690511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,32,1,power_law_1.01,0.1456704020500183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,32,balanced,0.05175999800364176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,32,balanced,0.05199466645717621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,32,balanced,0.06648533542950948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,32,1,power_law_1.01,0.149126398563385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,32,balanced,0.09708799918492635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,32,balanced,0.18020800749460855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,32,1,power_law_1.01,0.15102720260620117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,32,balanced,0.24949334065119425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,32,balanced,0.2547253370285034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,32,1,power_law_1.01,0.1580415964126587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,32,balanced,0.23106133937835693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,32,balanced,0.2358506719271342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,32,1,power_law_1.01,0.1665984034538269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,32,balanced,0.3367520173390706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,32,balanced,0.34756799538930255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,32,1,power_law_1.2,1.3113408088684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,32,1,power_law_1.01,0.17480319738388062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,32,balanced,0.35364798704783124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,32,balanced,0.03982933362325033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,32,balanced,0.041482667128245033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,32,balanced,0.36534400780995685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,32,1,power_law_1.01,0.18955520391464234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,32,balanced,0.039674667020638786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,32,balanced,0.05372266471385956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,32,balanced,0.37514666716257733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,32,power_law_1.2,2.069919967651367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,32,balanced,0.07038400073846181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,32,1,power_law_1.2,0.04341759979724884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,32,balanced,0.10742933551470439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,32,1,power_law_1.01,0.27068800926208497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,32,balanced,0.38601064682006836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,32,balanced,0.16267200311024985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,32,1,power_law_1.2,0.054713600873947145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,32,balanced,0.40937598546346027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,32,balanced,0.05826666454474131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,32,1,power_law_1.01,0.29656319618225097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,32,balanced,0.06033066908518473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,32,1,power_law_1.2,0.15559680461883546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,32,balanced,0.4720160166422526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,32,1,power_law_1.2,0.06731520295143127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,32,1,power_law_1.01,0.3465791940689087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,32,1,power_law_1.2,0.09277439713478089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,32,1,power_law_1.2,0.11592320203781128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,32,1,power_law_1.2,0.12684799432754518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,32,1,power_law_1.2,0.12956160306930542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,32,1,power_law_1.2,0.13058559894561766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,32,1,power_law_1.2,0.14669439792633057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,32,1,power_law_1.2,0.15020159482955933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,32,balanced,0.16332266728083292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,32,balanced,0.17083199818929037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,32,balanced,0.17511999607086182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,32,balanced,0.06021333237489065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,32,1,power_law_1.2,0.15337599515914918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,32,balanced,0.07666666805744171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,32,balanced,0.27911466360092163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,32,balanced,0.0886346697807312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,32,balanced,0.5042773485183716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,32,balanced,0.2919680078824361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,32,1,power_law_1.2,0.16191999912261962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,32,balanced,0.29803733030955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,32,balanced,0.5638720194498698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,32,balanced,0.3065173427263896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,32,1,power_law_1.01,0.39648640155792236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,32,balanced,0.32017600536346436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,32,balanced,0.624234676361084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,32,balanced,0.33110400040944415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,32,balanced,0.3543786605199178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,32,1,power_law_1.01,0.5179711818695069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,32,balanced,0.40161601702372235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,32,balanced,0.43389864762624103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,32,1,power_law_1.01,0.6327936172485351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,32,balanced,0.50381867090861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,32,balanced,0.5675786733627319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,32,1,power_law_1.01,0.8613439559936523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,32,balanced,0.7028106848398844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,32,power_law_1.2,2.90832633972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,32,balanced,0.8389493624369303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,32,1,power_law_1.2,0.16250879764556886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,32,balanced,0.12038399775822957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,32,balanced,0.14826132853825888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,32,1,power_law_1.01,1.1102208137512206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,32,balanced,0.14985066652297974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,32,1,power_law_1.2,0.1689919948577881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,32,balanced,1.1094720363616943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,32,balanced,0.3343573411305745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,32,balanced,0.33984001477559406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,32,1,power_law_1.2,0.17711999416351318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,32,balanced,0.7483680248260498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,32,balanced,1.4373760223388672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,32,1,power_law_1.2,0.19155839681625367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,32,1,power_law_1.01,1.5686976432800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,32,balanced,0.8716213703155518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,32,1,power_law_1.2,0.276147198677063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,32,balanced,1.1169119675954182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,32,1,power_law_1.2,0.2996543884277344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,32,1,power_law_1.2,0.3517312049865723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,32,balanced,1.463370641072591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,32,1,power_law_1.01,2.0481599807739257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,32,1,power_law_1.2,0.40380158424377444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,32,1,power_law_1.2,0.15326720476150513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,32,1,power_law_1.2,0.1552896022796631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,32,balanced,1.9664853413899739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,32,1,power_law_1.2,0.5250239849090577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,32,balanced,0.3457653522491455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,32,balanced,0.3535039822260539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,32,1,power_law_1.2,0.6382719993591308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,32,balanced,0.3607306480407715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,32,balanced,0.3742400010426839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,32,balanced,2.575391928354899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,32,balanced,0.389903982480367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,32,balanced,1.9787999788920085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,32,1,power_law_1.2,0.8670720100402832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,32,balanced,0.4025333325068156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,32,power_law_1.01,0.1507200002670288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,32,balanced,0.43080000082651776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,32,power_law_1.01,0.18442879915237426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,32,balanced,0.4794613520304362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,32,1,power_law_1.2,1.1099648475646973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,32,balanced,2.5823465983072915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,32,balanced,0.5309280157089233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,32,power_law_1.01,0.1828927993774414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,32,balanced,0.6295626560846964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,32,power_law_1.01,0.0871616005897522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,32,1,power_law_1.2,0.17748479843139647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,32,balanced,0.7314186890920004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,32,power_law_1.01,0.18484480381011964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,32,power_law_1.01,0.12522239685058595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,32,1,power_law_1.2,0.17831039428710938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,32,balanced,0.9310239950815836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,32,power_law_1.01,0.19104000329971313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,32,1,power_law_1.2,0.17828480005264283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,32,power_law_1.01,0.13352320194244385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,32,1,power_law_1.2,0.18880640268325805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,32,balanced,1.2409919897715251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,32,power_law_1.01,0.1987264037132263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,32,power_law_1.01,0.13473279476165773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,32,1,power_law_1.2,0.1967743992805481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,32,1,power_law_1.2,0.20330240726470947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,32,balanced,1.64082670211792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,32,1,power_law_1.2,0.21765120029449464
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,32,power_law_1.2,0.5626815795898438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,32,balanced,2.152106602986654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,32,balanced,0.03143999973932902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,32,balanced,0.03133333226044973
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,32,balanced,0.03148799886306127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,32,balanced,0.0322773332397143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,32,balanced,3.061333338419596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,32,power_law_1.01,0.214355206489563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,32,power_law_1.01,0.13521920442581176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,32,power_law_1.01,0.2078847885131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,32,1,power_law_1.2,0.32223360538482665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,32,balanced,4.07532278696696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,32,power_law_1.01,0.20906240940093995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,32,1,power_law_1.2,0.35054080486297606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,32,power_law_1.01,0.18942079544067383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,32,1,power_law_1.2,0.4006336212158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,32,power_law_1.01,0.10016640424728393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,32,balanced,0.04178666571776072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,32,power_law_1.01,0.315065598487854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,32,balanced,0.06186666587988535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,32,balanced,0.07825600107510884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,32,balanced,0.08061333497365315
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,32,balanced,0.08091199894746144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,32,power_law_1.01,0.3273279905319214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,32,balanced,0.08271466692288716
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,32,balanced,0.12168000141779582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,32,balanced,0.11309333642323811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,32,power_law_1.01,0.3184256076812744
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,32,balanced,0.11076800028483073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,32,power_law_1.01,0.1204800009727478
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,32,balanced,0.2687573234240214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,32,balanced,0.2141546607017517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,32,power_law_1.01,0.343724799156189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,32,power_law_1.01,0.14865280389785768
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,32,balanced,0.18181333939234415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,32,balanced,0.1104853351910909
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,32,power_law_1.01,0.3467072010040283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,32,power_law_1.01,0.15564800500869752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,32,power_law_1.01,0.3580480098724365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,32,1,power_law_1.2,0.4493120193481445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,32,power_law_1.01,0.13204480409622193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,32,power_law_1.01,0.1421183943748474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,32,power_law_1.2,0.8211263656616211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,32,1,power_law_1.2,0.5873472213745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,32,power_law_1.01,0.15062400102615356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,32,power_law_1.01,0.13676799535751344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,32,power_law_1.01,0.26102399826049805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,32,1,power_law_1.2,0.6959871768951416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,32,power_law_1.01,0.27486720085144045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,32,balanced,0.11752532919247945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,32,power_law_1.01,0.2856064081192017
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,32,balanced,0.1256480018297831
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,32,balanced,0.14511467019716898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,32,power_law_1.01,0.2918272018432617
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,32,balanced,0.16667733589808145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,32,balanced,0.2075786590576172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,32,power_law_1.01,0.300217604637146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,32,balanced,0.25044800837834674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,32,power_law_1.01,0.40152320861816404
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,32,balanced,0.33446399370829266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,32,power_law_1.01,0.3165503978729248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,32,balanced,0.4392000039418538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,32,power_law_1.01,0.4673408031463623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,32,power_law_1.01,0.13777920007705688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,32,power_law_1.01,0.3350080013275146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,32,balanced,0.593946655591329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,32,power_law_1.01,0.48944640159606934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,32,power_law_1.01,0.13848960399627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,32,power_law_1.01,0.3878976106643677
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,32,balanced,0.8104106585184733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,32,power_law_1.01,0.5802495956420899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,32,power_law_1.01,0.13752319812774658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,32,power_law_1.01,0.4356095790863037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,32,1,power_law_1.2,0.9442048072814941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,32,power_law_1.2,3.6949249267578126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,32,power_law_1.01,0.6479743957519531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,32,power_law_1.01,0.5057536125183105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,32,power_law_1.2,0.1490623950958252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,32,power_law_1.01,0.5864960193634033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,32,power_law_1.01,0.8143232345581055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,32,power_law_1.2,0.18199679851531983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,32,power_law_1.01,0.7559167861938476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,32,power_law_1.2,0.18529280424118041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,32,power_law_1.01,0.9802623748779297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,32,power_law_1.2,0.18492159843444825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,32,power_law_1.01,0.8983424186706543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,32,power_law_1.01,1.3066816329956055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,32,power_law_1.2,0.19123200178146363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,32,power_law_1.01,0.1377087950706482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,32,1,power_law_1.2,1.5858943939208985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,32,power_law_1.01,1.2333760261535645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,32,power_law_1.2,0.20725760459899903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,32,power_law_1.01,0.1433727979660034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,32,power_law_1.01,1.6697023391723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,32,power_law_1.2,0.1928256034851074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,32,power_law_1.01,1.549343967437744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,32,power_law_1.01,0.3255743980407715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,32,1,power_law_1.2,1.2045120239257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,32,power_law_1.2,0.21986560821533202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,32,power_law_1.01,0.30770559310913087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,32,power_law_1.01,2.414361572265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,32,power_law_1.01,0.32583041191101075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,32,power_law_1.01,2.168320083618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,32,power_law_1.01,0.34374399185180665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,32,power_law_1.01,0.3518527984619141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,32,power_law_1.01,3.0549312591552735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,32,power_law_1.01,0.35392639636993406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,32,power_law_1.01,2.826508712768555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,32,power_law_1.2,0.08706560134887695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,32,power_law_1.01,0.3787519931793213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,32,power_law_1.2,0.12744959592819213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,32,power_law_1.2,0.09905920028686524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,32,power_law_1.2,0.1831936001777649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,32,power_law_1.01,0.38798720836639405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,32,power_law_1.2,0.13061120510101318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,32,power_law_1.2,0.19560320377349855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,32,power_law_1.01,0.4202432155609131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,32,power_law_1.2,0.13608319759368898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,32,power_law_1.2,0.30091519355773927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,32,power_law_1.01,0.48784642219543456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,32,power_law_1.2,0.1356735944747925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,32,power_law_1.2,0.3155711889266968
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,32,power_law_1.01,0.05578240156173706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,32,power_law_1.2,0.1384703993797302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,32,power_law_1.01,0.5590015888214112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,32,power_law_1.01,0.06904320120811462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,32,power_law_1.2,0.309004807472229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,32,power_law_1.2,0.14095360040664673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,32,power_law_1.01,0.7033792018890381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,32,power_law_1.01,0.06872320175170898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,32,power_law_1.2,0.3359616041183472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,32,power_law_1.2,0.13451520204544068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,32,power_law_1.2,0.13623679876327516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,32,power_law_1.01,0.06926720142364502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,32,power_law_1.2,0.3488895893096924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,32,power_law_1.01,0.8449472427368164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,32,power_law_1.2,0.13557759523391724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,32,power_law_1.2,0.13515520095825195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,32,power_law_1.01,0.07230079770088196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,32,power_law_1.2,0.3557248115539551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,32,power_law_1.2,0.13742719888687133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,32,power_law_1.01,1.1336383819580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,32,power_law_1.2,0.31159679889678954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,32,power_law_1.01,0.07178239822387696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,32,power_law_1.2,0.38536319732666013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,32,power_law_1.2,0.13281279802322388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,32,power_law_1.2,0.3227519989013672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,32,power_law_1.2,0.4662335872650146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,32,power_law_1.01,1.4121919631958009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,32,power_law_1.2,0.12803839445114135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,32,power_law_1.2,0.31790719032287595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,32,power_law_1.2,0.500486421585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,32,power_law_1.2,0.14928640127182008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,32,power_law_1.2,0.33803520202636717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,32,power_law_1.2,0.5994815826416016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,32,power_law_1.01,2.0523136138916014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,32,power_law_1.2,0.14958720207214354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,32,power_law_1.2,0.3391103982925415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,32,power_law_1.2,0.6834943771362305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,32,power_law_1.2,0.15354880094528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,32,power_law_1.2,0.3554879903793335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,32,power_law_1.2,0.8456319808959961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,32,power_law_1.2,0.26136319637298583
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,32,power_law_1.01,0.0673919975757599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,32,power_law_1.2,0.3694272041320801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,32,power_law_1.01,2.5660736083984377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,32,power_law_1.2,0.2708031892776489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,32,1,power_law_1.2,1.698259162902832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,32,power_law_1.01,0.07464320063591004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,32,power_law_1.2,0.3911360025405884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,32,power_law_1.2,0.27619199752807616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,32,power_law_1.01,0.07728000283241272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,32,power_law_1.2,0.4234943866729736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,32,power_law_1.2,0.2841536045074463
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,32,power_law_1.01,0.07521920204162598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,32,power_law_1.2,0.5032959938049316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,32,power_law_1.2,0.31086719036102295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,32,power_law_1.01,0.10831359624862671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,32,power_law_1.2,0.5617663860321045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,32,power_law_1.01,3.789267349243164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,32,power_law_1.2,0.3099456071853638
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,32,power_law_1.01,0.092985600233078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,32,power_law_1.2,0.7329984188079834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,32,power_law_1.2,0.3474047899246216
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,32,power_law_1.01,0.0945855975151062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,32,power_law_1.2,1.0504063606262206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,32,power_law_1.2,0.8777152061462402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,32,power_law_1.01,0.2218303918838501
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,32,power_law_1.01,0.17214080095291137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,32,power_law_1.2,1.435110378265381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,32,power_law_1.2,1.1910335540771484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,32,power_law_1.01,0.14217599630355834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,32,power_law_1.01,4.975859069824219
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,32,power_law_1.01,0.10395519733428955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,32,power_law_1.2,1.5363776206970214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,32,power_law_1.2,1.8265983581542968
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,32,power_law_1.01,0.11056640148162841
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,32,power_law_1.01,0.12535680532455445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,32,power_law_1.2,0.3949824094772339
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,32,power_law_1.01,0.14706560373306274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,32,power_law_1.2,0.43312640190124513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,32,power_law_1.2,2.222035217285156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,32,power_law_1.2,2.493401527404785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,32,power_law_1.2,0.5338816165924072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,32,power_law_1.2,0.5980095863342285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,32,power_law_1.2,2.858348846435547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,32,power_law_1.2,0.7691775798797608
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,32,power_law_1.2,1.1285823822021483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,32,power_law_1.2,3.6766014099121094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,32,1,balanced,0.03988266736268997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,32,1,balanced,0.031514666974544525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,32,1,balanced,0.0356480007370313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,32,1,balanced,0.043765331308046974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,32,1,balanced,0.05956799785296122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,32,1,balanced,0.09935466448465984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,32,power_law_1.01,0.17111680507659913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,32,1,balanced,0.1609546641508738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,32,1,balanced,0.27137066920598346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,32,1,balanced,0.27365867296854657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,32,power_law_1.01,0.21634559631347655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,32,1,balanced,0.25057599941889447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,32,1,balanced,0.2527573307355245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,32,power_law_1.2,3.9399742126464843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,32,1,balanced,0.31381332874298096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,32,power_law_1.01,0.26968319416046144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,32,1,balanced,0.3144320050875346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,32,1,balanced,0.3184746702512105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,32,1,balanced,0.3253119985262553
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,32,power_law_1.01,0.3794624090194702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,32,1,balanced,0.33215999603271484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,32,1,balanced,0.33854401111602783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,32,1,balanced,0.35073598225911456
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,32,power_law_1.01,0.4784575939178467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,32,1,balanced,0.5074346860249838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,32,power_law_1.2,0.9267647743225098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,32,1,balanced,0.5289706786473592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,32,1,balanced,0.5668319861094157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,32,power_law_1.01,0.6998720169067383
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,32,power_law_1.2,0.05740159749984741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,32,1,balanced,0.6076106627782186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,32,power_law_1.2,1.2874815940856934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,32,1,balanced,0.6931146780649821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,32,power_law_1.2,5.4227455139160154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,32,1,power_law_1.2,2.065235137939453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,32,1,balanced,0.7753439744313558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,32,power_law_1.01,0.9911487579345704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,32,1,power_law_1.01,0.03746559917926788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,32,1,power_law_1.2,2.207961654663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,32,1,balanced,0.9450399875640869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,32,power_law_1.2,1.6335039138793945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,32,1,power_law_1.01,0.044537600874900815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,32,1,balanced,0.04588800172011057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,32,1,balanced,1.3431946436564128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,32,1,balanced,0.06211733321348826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,32,1,balanced,0.09363200267155965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,32,1,power_law_1.01,0.05994880199432373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,32,1,balanced,0.16835200786590576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,32,1,balanced,0.3261066675186157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,32,1,balanced,0.43383467197418213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,32,1,balanced,1.671354611714681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,32,1,balanced,0.4389866590499878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,32,power_law_1.2,2.3199167251586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,32,1,balanced,0.2815893292427063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,32,power_law_1.2,0.07021440267562866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,32,1,balanced,0.2835306723912557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,32,1,balanced,0.353765328725179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,32,1,balanced,0.3529333273569743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,32,power_law_1.2,0.07095680236816407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,32,1,balanced,0.3553760051727295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,32,1,balanced,0.357968012491862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,32,1,power_law_1.01,0.03000960052013397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,32,1,balanced,0.3659679889678955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,32,power_law_1.2,0.0704576015472412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,32,1,balanced,0.3703999916712443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,32,1,power_law_1.01,0.033395200967788696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,32,1,balanced,0.38182934125264484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,32,power_law_1.2,3.0089792251586913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,32,power_law_1.2,0.07296640276908875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,32,1,balanced,0.5577173233032227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,32,1,power_law_1.01,0.03937920033931732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,32,1,balanced,0.5764373143513998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,32,1,power_law_1.01,0.09256319999694824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,32,1,balanced,0.6118613481521606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,32,1,power_law_1.01,0.06117759943008423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,32,1,balanced,0.647434671719869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,32,1,power_law_1.01,0.0908352017402649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,32,1,power_law_1.01,0.15724799633026124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,32,1,balanced,0.72652800877889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,32,1,balanced,2.239408016204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,32,1,balanced,0.8062026500701904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,32,1,power_law_1.01,0.11969280242919922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,32,1,power_law_1.01,0.2169343948364258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,32,1,power_law_1.2,0.029791998863220214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,32,1,balanced,0.9667999744415283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,32,1,power_law_1.01,0.17395199537277223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,32,1,power_law_1.01,0.23995521068572997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,32,1,power_law_1.2,0.033523198962211606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,32,1,balanced,1.3604906400044758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,32,1,power_law_1.01,0.20604159832000732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,32,1,power_law_1.2,0.044614401459693906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,32,1,power_law_1.01,0.2885119915008545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,32,1,power_law_1.2,0.03736959993839264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,32,1,power_law_1.01,0.19321600198745728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,32,power_law_1.2,0.07368320226669312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,32,1,balanced,1.6744853655497234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,32,1,power_law_1.2,0.044563201069831845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,32,1,power_law_1.01,0.21080319881439208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,32,1,power_law_1.01,0.02919679880142212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,32,1,power_law_1.01,0.20268800258636474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,32,1,power_law_1.2,0.06229760050773621
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,32,power_law_1.2,0.07294080257415772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,32,1,power_law_1.01,0.22684159278869628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,32,1,power_law_1.01,0.030393600463867188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,32,1,power_law_1.01,0.27073280811309813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,32,1,balanced,2.2433600425720215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,32,1,power_law_1.2,0.09269760251045227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,32,power_law_1.2,0.073990398645401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,32,1,power_law_1.01,0.3051136016845703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,32,1,power_law_1.01,0.03324800133705139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,32,1,power_law_1.2,0.14049279689788818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,32,1,power_law_1.01,0.3035648107528687
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,32,power_law_1.2,0.07091839909553528
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,32,1,balanced,0.02922133356332779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,32,1,power_law_1.2,0.1955135941505432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,32,1,balanced,0.03129599988460541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,32,1,power_law_1.01,0.3088896036148071
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,32,power_law_1.2,0.07735040187835693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,32,1,power_law_1.2,0.21944959163665773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,32,1,power_law_1.2,0.06101120114326477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,32,1,power_law_1.01,0.3210367918014526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,32,power_law_1.2,0.09903360009193421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,32,1,power_law_1.2,0.08650239706039428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,32,1,power_law_1.01,0.33045120239257814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,32,1,power_law_1.2,0.27791359424591067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,32,power_law_1.2,0.09934080243110657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,32,1,power_law_1.01,0.33588480949401855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,32,1,power_law_1.2,0.11202559471130372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,32,1,power_law_1.2,0.20929279327392578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,32,1,power_law_1.01,0.2781183958053589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,32,1,power_law_1.01,0.35155200958251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,32,1,power_law_1.2,0.16090879440307618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,32,1,power_law_1.2,0.21502079963684081
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,32,1,power_law_1.01,0.041094401478767396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,32,1,power_law_1.01,0.2818880081176758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,32,1,power_law_1.01,0.5208831787109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,32,1,power_law_1.01,0.28850560188293456
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,32,1,balanced,0.033402666449546814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,32,1,power_law_1.2,0.20007040500640869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,32,1,power_law_1.2,0.3009344100952148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,32,1,balanced,0.03974399964014689
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,32,1,balanced,0.05765866736570994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,32,1,balanced,0.09683733185132344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,32,1,power_law_1.01,0.30097920894622804
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,32,1,balanced,0.13127467036247253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,32,1,power_law_1.2,0.1850816011428833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,32,1,power_law_1.2,0.3004928112030029
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,32,1,balanced,0.13338133692741394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,32,1,balanced,0.13597866892814636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,32,1,power_law_1.01,0.30944640636444093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,32,1,balanced,0.13869866728782654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,32,1,power_law_1.2,0.19292800426483153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,32,1,power_law_1.2,0.3097471952438354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,32,1,power_law_1.01,0.3239487886428833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,32,power_law_1.2,0.0923520028591156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,32,1,power_law_1.2,0.2682816028594971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,32,1,power_law_1.01,0.47000961303710936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,32,power_law_1.2,0.21130878925323487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,32,1,power_law_1.2,0.2754751920700073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,32,1,power_law_1.01,0.49370241165161133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,32,power_law_1.2,0.1500159978866577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,32,1,power_law_1.2,0.2809727907180786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,32,1,power_law_1.01,0.550867223739624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,32,1,power_law_1.2,0.028454399108886717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,32,1,power_law_1.01,0.5448256015777588
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,32,power_law_1.2,0.15141760110855101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,32,1,power_law_1.2,0.28781440258026125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,32,1,power_law_1.2,0.030086401104927062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,32,1,power_law_1.01,0.6109439849853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,32,1,power_law_1.01,0.5956223964691162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,32,power_law_1.2,0.10247039794921875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,32,1,balanced,0.14256532986958823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,32,1,power_law_1.2,0.29923200607299805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,32,1,balanced,0.14731199542681375
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,32,1,power_law_1.2,0.03311359882354736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,32,1,balanced,0.15425067146619162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,32,power_law_1.2,0.11516799926757812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,32,1,power_law_1.01,0.7252543926239013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,32,1,power_law_1.2,0.30743041038513186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,32,1,power_law_1.01,0.6517759799957276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,32,1,power_law_1.2,0.0411327987909317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,32,power_law_1.2,0.12934399843215943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,32,1,power_law_1.2,0.3287616014480591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,32,1,power_law_1.01,0.8424063682556152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,32,1,power_law_1.01,0.7557888031005859
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,32,1,power_law_1.2,0.054995197057724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,32,power_law_1.2,0.14935679435729982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,32,1,power_law_1.2,0.4712063789367676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,32,1,power_law_1.01,0.05653759837150574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,32,power_law_1.2,0.17820800542831422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,32,1,power_law_1.2,0.08042240142822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,32,1,power_law_1.2,0.5075712203979492
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,32,1,power_law_1.01,0.08359040021896362
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,32,power_law_1.2,0.24324479103088378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,32,1,power_law_1.2,0.1007807970046997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,32,1,power_law_1.01,0.10304640531539917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,32,1,power_law_1.2,0.5666751861572266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,32,power_law_1.2,0.3101759910583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,32,1,power_law_1.2,0.11471999883651733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,32,1,power_law_1.01,0.11427839994430541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,32,1,power_law_1.2,0.6256703853607177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,32,1,power_law_1.2,0.31910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,32,power_law_1.2,0.4216832160949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,32,1,power_law_1.2,0.11783679723739623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,32,1,power_law_1.2,0.32877440452575685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,32,1,power_law_1.2,0.7410816192626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,32,1,power_law_1.2,0.33678081035614016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,32,power_law_1.2,0.5617792129516601
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,32,1,power_law_1.2,0.12177280187606812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,32,1,power_law_1.2,0.8666943550109864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,32,1,power_law_1.01,0.8608768463134766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,32,1,power_law_1.01,1.0985024452209473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,32,1,power_law_1.2,0.35625600814819336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,32,1,power_law_1.2,0.1300287961959839
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,32,power_law_1.2,0.7574143886566163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,32,1,power_law_1.2,1.1275775909423829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,32,1,power_law_1.2,0.5237184047698975
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,32,1,power_law_1.2,0.13738240003585817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,32,1,power_law_1.01,1.129695987701416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,32,1,power_law_1.01,1.33755521774292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,32,1,balanced,0.16264533003171286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,32,1,balanced,0.17326400677363077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,32,1,power_law_1.2,0.5565248012542725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,32,1,balanced,0.1832266648610433
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,32,1,balanced,0.22858667373657227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,32,1,power_law_1.01,0.11755520105361938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,32,1,power_law_1.2,1.3616319656372071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,32,1,power_law_1.01,1.3522879600524902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,32,1,balanced,0.2701920072237651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,32,1,power_law_1.2,0.614246416091919
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,32,1,balanced,0.25806933641433716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,32,1,power_law_1.01,1.8362943649291992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,32,1,power_law_1.01,0.12343039512634277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,32,1,power_law_1.2,0.6671167850494385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,32,1,power_law_1.01,0.13048319816589354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,32,1,power_law_1.2,1.8620159149169921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,32,1,power_law_1.01,1.8526464462280274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,32,1,power_law_1.2,0.7743936061859131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,32,1,power_law_1.01,0.1377344012260437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,32,1,power_law_1.01,2.3281856536865235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,32,1,power_law_1.01,0.14609919786453246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,balanced,0.033226666351159416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,32,1,power_law_1.2,0.14643839597702027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,balanced,0.03342399994532267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,balanced,0.03345066557327906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,32,power_law_1.2,1.1176063537597656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,balanced,0.035242666800816856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,32,1,power_law_1.01,0.1598528027534485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,balanced,0.047695999344189964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,balanced,0.03155199935038885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,balanced,0.07818666597207387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,balanced,0.03161599983771642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,balanced,0.07597866654396057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,balanced,0.03139200061559677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,32,1,power_law_1.2,2.363180732727051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,balanced,0.07587733368078868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,balanced,0.03363733241955439
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,32,1,power_law_1.01,0.17342720031738282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,balanced,0.08094933132330577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,balanced,0.03773866593837738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,balanced,0.0825973351796468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,balanced,0.055904000997543335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,balanced,0.10274133086204529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,balanced,0.06852266689141591
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,32,1,balanced,0.3131733338038127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,balanced,0.10928533474604289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,balanced,0.06875733534495036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,32,1,power_law_1.01,0.1894144058227539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,balanced,0.1123253305753072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,balanced,0.06884799897670746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,32,1,balanced,0.3237226605415344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,balanced,0.11719466249148051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,balanced,0.12356266379356384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,32,1,balanced,0.40514667828877765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,balanced,0.12844266494115195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,32,1,power_law_1.01,0.21427841186523439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,balanced,0.1399733324845632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,32,1,balanced,0.5481760104497274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,balanced,0.17244799931844076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,balanced,0.1894986629486084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,32,1,power_law_1.01,0.27628159523010254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,32,1,balanced,0.6810080210367838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,32,1,power_law_1.01,0.24825599193572997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,32,1,balanced,0.9266560077667236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,32,1,power_law_1.2,0.8858240127563477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,32,1,power_law_1.01,0.3208575963973999
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,32,1,power_law_1.2,0.16268800497055053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,32,1,balanced,0.9945120016733805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,32,1,power_law_1.01,2.358220863342285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,32,1,power_law_1.01,0.31637120246887207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,32,1,power_law_1.2,1.1547648429870605
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,32,1,power_law_1.2,0.17538559436798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,32,1,balanced,1.3474292755126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,balanced,0.06818133095900218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,32,1,power_law_1.01,0.4204160213470459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,balanced,0.101200004418691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,balanced,0.10736533006032307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,power_law_1.01,0.05985919833183288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,balanced,0.11102933684984843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,balanced,0.11635733644167583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,balanced,0.12479999661445618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,32,1,power_law_1.01,0.5521984100341797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,32,1,power_law_1.2,1.3766207695007324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,power_law_1.01,0.05948160290718078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,balanced,0.22405334313710532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,balanced,0.2587520082791646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,power_law_1.01,0.05939840078353882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,balanced,0.32740267117818195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,32,1,power_law_1.01,0.7525504112243653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,power_law_1.01,0.06043519973754883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,balanced,0.39422933260599774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,balanced,0.5583626826604208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,power_law_1.01,0.06020479798316956
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,32,1,power_law_1.01,0.7758272171020508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,balanced,0.6937440236409506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,power_law_1.01,0.06346240043640136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,power_law_1.01,0.06885759830474854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,32,1,power_law_1.2,0.18869760036468505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,32,1,power_law_1.01,1.066528034210205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,power_law_1.01,0.0699455976486206
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,32,1,power_law_1.2,0.21749119758605956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,balanced,0.1295146644115448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,power_law_1.01,0.08028159737586975
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,balanced,0.027386667827765148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,balanced,0.13981333374977112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,balanced,0.027621333797772724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,balanced,0.16621333360671997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,power_law_1.01,0.07973120212554932
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,32,1,power_law_1.2,0.2785856008529663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,balanced,0.1844373345375061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,balanced,0.22588799397150675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,power_law_1.01,0.09830399751663207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,balanced,0.2646720012029012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,balanced,0.34201598167419434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,power_law_1.01,0.103603196144104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,32,1,power_law_1.2,1.8786624908447265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,balanced,0.42162132263183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,power_law_1.01,0.10805759429931641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,balanced,0.03990400085846583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,balanced,0.04195733368396759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,balanced,0.5888426701227824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,balanced,0.0421013335386912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,power_law_1.01,0.11233279705047608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,balanced,0.04408533374468485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,balanced,0.04809066653251648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,balanced,0.9930666287740072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,balanced,0.741866668065389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,balanced,0.06875200072924297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,power_law_1.01,0.12008960247039795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,power_law_1.01,0.12730879783630372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,balanced,1.06440003712972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,balanced,1.286186695098877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,32,1,power_law_1.2,2.380147171020508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,power_law_1.01,0.13833600282669067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,power_law_1.01,0.04832000136375427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,balanced,0.02741866558790207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,balanced,1.3815520604451497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,power_law_1.01,0.1764032006263733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,power_law_1.01,0.05137280225753784
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,32,1,power_law_1.01,1.4296064376831055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,32,1,power_law_1.2,0.2519551992416382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,power_law_1.01,0.1966271996498108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,power_law_1.2,0.06044160127639771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,power_law_1.01,0.24330239295959472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,32,1,power_law_1.2,0.32254719734191895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,power_law_1.2,0.06037120223045349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,power_law_1.01,0.28670079708099366
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,32,1,power_law_1.2,0.3210560083389282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,power_law_1.2,0.060102397203445436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,power_law_1.01,0.3685375928878784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,balanced,0.0709440012772878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,power_law_1.2,0.060736000537872314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,32,1,power_law_1.2,0.4234943866729736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,power_law_1.01,0.4492032051086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,power_law_1.2,0.05940480232238769
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,32,1,power_law_1.2,0.5574975967407226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,power_law_1.2,0.06474879980087281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,balanced,0.02735466758410136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,balanced,0.02941333254178365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,balanced,0.03366400053103765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,power_law_1.2,0.0713599979877472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,32,1,power_law_1.2,0.7654911994934082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,balanced,0.03598399957021078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,balanced,0.03775466730197271
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,balanced,0.03968533376852671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,balanced,0.03995199998219808
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,balanced,0.05073066552480062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,balanced,0.09974933664004008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,32,1,power_law_1.2,0.7868480205535888
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,balanced,0.08741333087285359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,balanced,0.07935466865698497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,balanced,0.06162666777769724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,balanced,0.058431997895240784
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,balanced,0.05815466741720835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,power_law_1.01,0.029971200227737426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,balanced,0.0565226674079895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,balanced,0.06019733349482218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,32,1,power_law_1.2,1.0800000190734864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,balanced,0.06875733534495036
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,power_law_1.01,0.030118399858474733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,power_law_1.01,0.627839994430542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,power_law_1.01,0.029606398940086365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,power_law_1.01,0.02963840067386627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,power_law_1.01,0.05188480019569397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,power_law_1.01,0.8063808441162109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,power_law_1.2,0.06755200028419495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,power_law_1.01,0.030163198709487915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,power_law_1.01,0.05146239995956421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,power_law_1.2,0.07973759770393371
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,power_law_1.01,0.032025599479675294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,power_law_1.01,0.04907520115375519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,power_law_1.01,1.1477375984191895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,power_law_1.2,0.08024320006370544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,balanced,0.07281599938869476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,power_law_1.01,0.034457600116729735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,balanced,0.11989333232243855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,balanced,0.12197867035865784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,balanced,0.12471999724706014
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,balanced,0.0806826651096344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,power_law_1.01,0.0358271986246109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,balanced,0.12982400258382162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,balanced,0.10528000195821126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,balanced,0.13250133395195007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,balanced,0.14232533176740012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,power_law_1.01,0.05846400260925293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,power_law_1.01,0.03844479918479919
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,power_law_1.01,0.03917439877986908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,power_law_1.01,0.05902720093727112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,32,1,power_law_1.2,1.4475775718688966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,power_law_1.01,0.045465600490570066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,power_law_1.2,0.04878720045089722
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,power_law_1.01,0.0828224003314972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,power_law_1.2,0.05214080214500427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,power_law_1.01,0.07946879863739013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,power_law_1.2,0.050400000810623166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,power_law_1.01,0.05120000243186951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,power_law_1.2,0.09866880178451538
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,power_law_1.01,0.06711040139198303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,power_law_1.2,0.05162879824638367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,power_law_1.01,0.06218879818916321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,power_law_1.2,0.10387840270996093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,power_law_1.01,0.05443840026855469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,power_law_1.01,1.459552001953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,power_law_1.01,0.06254079937934875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,power_law_1.2,0.049856001138687135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,power_law_1.2,0.10737919807434082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,balanced,0.12575999895731607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,balanced,0.1495786706606547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,power_law_1.01,0.05348479747772217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,power_law_1.01,0.06581119894981384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,balanced,0.1558613379796346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,power_law_1.2,0.05799040198326111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,power_law_1.2,0.11334400177001953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,power_law_1.2,0.051596802473068235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,balanced,0.17251733938852945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,power_law_1.01,0.05464320182800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,power_law_1.01,0.0665727972984314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,power_law_1.2,0.05886080265045166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,power_law_1.2,0.060115200281143186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,power_law_1.01,0.09894400238990783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,power_law_1.01,0.058432000875473025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,power_law_1.2,0.058259201049804685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,power_law_1.2,0.06304640173912049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,power_law_1.01,0.10572799444198608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,power_law_1.2,0.06467840075492859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,power_law_1.01,0.10817919969558716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,power_law_1.01,0.11379200220108032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,power_law_1.2,0.06528000235557556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,power_law_1.01,0.1211840033531189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,power_law_1.2,0.09747200012207032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,power_law_1.01,0.12785919904708862
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,balanced,0.17124799887339273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,power_law_1.2,0.10565119981765747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,balanced,0.2141973376274109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,power_law_1.01,0.1407487988471985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,balanced,0.20158400138219199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,power_law_1.2,0.12478079795837402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,balanced,0.22986133893330893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,power_law_1.01,0.16587519645690918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,power_law_1.2,0.12899199724197388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,power_law_1.01,0.05894399881362915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,power_law_1.01,0.1863103985786438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,power_law_1.2,0.14535679817199706
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,power_law_1.01,0.06387839913368225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,power_law_1.01,0.05894399881362915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,power_law_1.01,0.23841919898986816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,power_law_1.2,0.18064639568328858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,power_law_1.01,0.07805439829826355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,power_law_1.01,0.05904639959335327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,power_law_1.01,0.2817984104156494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,power_law_1.2,0.20410239696502686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,power_law_1.01,0.09041919708251953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,power_law_1.01,0.06272640228271484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,power_law_1.01,0.36574079990386965
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,power_law_1.01,0.11299200057983398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,power_law_1.01,0.06769279837608337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,power_law_1.01,0.4557759761810303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,power_law_1.01,0.13846399784088134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,power_law_1.01,0.0682752013206482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,power_law_1.2,0.10768640041351318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,balanced,0.3019146720568339
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,power_law_1.01,0.19196799993515015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,power_law_1.01,0.11646720170974731
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,balanced,0.3889919916788737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,power_law_1.2,0.11487360000610351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,balanced,0.2908426721890767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,power_law_1.01,0.11914880275726318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,power_law_1.2,0.1230463981628418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,power_law_1.2,0.05902079939842224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,power_law_1.2,0.12887040376663209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,power_law_1.01,0.12338559627532959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,power_law_1.2,0.056857597827911374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,power_law_1.2,0.14254080057144164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,power_law_1.01,0.12465280294418335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,power_law_1.2,0.06220800280570984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,power_law_1.2,0.16830719709396363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,power_law_1.01,0.12911360263824462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,power_law_1.2,0.24908161163330078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,power_law_1.2,0.19343359470367433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,power_law_1.2,0.06443520188331604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,power_law_1.01,0.13861119747161865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,power_law_1.2,0.2941632032394409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,power_law_1.2,0.23912959098815917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,power_law_1.2,0.06768640279769897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,power_law_1.01,0.6281216144561768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,power_law_1.2,0.393286395072937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,power_law_1.2,0.11502079963684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,power_law_1.2,0.2868031978607178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,balanced,0.3496053218841553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,power_law_1.2,0.11465599536895751
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,power_law_1.01,0.24778239727020263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,power_law_1.2,0.4856383800506592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,power_law_1.01,0.7994239807128907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,balanced,0.4952053229014079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,power_law_1.2,0.12083200216293336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,power_law_1.01,0.3373248100280762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,balanced,0.610586682955424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,power_law_1.2,0.6873663902282715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,power_law_1.2,0.1233024001121521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,power_law_1.01,1.1621312141418456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,balanced,0.8682666619618734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,power_law_1.2,0.12700159549713136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,power_law_1.01,0.46412158012390137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,power_law_1.2,0.8171327590942383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,balanced,0.025253333151340485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,power_law_1.2,0.03020800054073334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,power_law_1.2,0.13882240056991577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,balanced,1.1244800090789795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,power_law_1.2,0.14624639749526977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,power_law_1.01,0.14799360036849976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,power_law_1.01,1.497036838531494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,power_law_1.2,0.16176639795303344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,power_law_1.2,1.2318528175354004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,power_law_1.01,0.1537343978881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,balanced,1.6376427014668782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,power_law_1.2,0.17983360290527345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,power_law_1.01,0.17258880138397217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,power_law_1.2,0.37351040840148925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,power_law_1.2,0.21865599155426024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,power_law_1.01,0.2168639898300171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,balanced,2.1538880666097007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,power_law_1.2,1.578003215789795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,power_law_1.2,0.25759360790252683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,power_law_1.01,0.2487488031387329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,power_law_1.01,0.02234880030155182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,power_law_1.01,0.022515200078487396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,power_law_1.2,0.34119040966033937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,power_law_1.01,0.32505600452423095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,power_law_1.01,0.02645759880542755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,power_law_1.01,0.023468799889087677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,balanced,0.027600000301996868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,balanced,0.0337119996547699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,power_law_1.01,0.03345920145511627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,power_law_1.01,0.3994175910949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,power_law_1.2,0.42291841506958006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,power_law_1.01,0.026495999097824095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,power_law_1.2,0.02951039969921112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,power_law_1.01,0.046342399716377256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,balanced,0.023024000227451324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,power_law_1.01,0.033817601203918454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,balanced,0.023354666928450268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,power_law_1.2,0.5835008144378662
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,power_law_1.2,0.029017600417137145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,balanced,0.028090665737787884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,balanced,0.0331839993596077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,power_law_1.01,0.06749439835548401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,balanced,0.05000533163547516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,balanced,0.08412266770998637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,balanced,0.10136000315348308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,power_law_1.2,0.029292801022529603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,balanced,0.0997759997844696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,power_law_1.01,0.0889855980873108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,power_law_1.2,0.7642879962921143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,balanced,0.09655466675758362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,balanced,0.09614933530489604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,balanced,0.11225600043932597
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,power_law_1.2,0.029574400186538695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,power_law_1.2,0.4619647979736328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,power_law_1.01,0.10663679838180543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,balanced,0.11341333389282227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,balanced,0.11236266295115153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,balanced,0.11843732992808025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,power_law_1.2,1.088486385345459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,power_law_1.01,0.10802559852600098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,power_law_1.2,0.6484672069549561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,power_law_1.01,0.08773760199546814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,balanced,0.04568000137805939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,balanced,0.0804906686147054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,power_law_1.01,0.0889024019241333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,power_law_1.2,0.8465279579162598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,power_law_1.2,1.3761088371276855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,power_law_1.01,0.11201920509338378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,power_law_1.01,0.5513855934143066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,power_law_1.01,0.11129599809646606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,power_law_1.01,0.04595839977264404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,power_law_1.2,1.2087936401367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,power_law_1.01,0.11290880441665649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,power_law_1.01,0.7000895977020264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,power_law_1.01,0.057734400033950806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,power_law_1.2,2.1190464019775392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,power_law_1.01,0.11767040491104126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,power_law_1.01,0.07978879809379577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,power_law_1.01,0.12199039459228515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,power_law_1.01,1.005561637878418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,balanced,0.12158399820327759
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,power_law_1.2,0.03059200048446655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,power_law_1.01,0.08243200182914734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,balanced,0.12494400143623352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,power_law_1.01,0.12383999824523925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,balanced,0.13115200400352478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,balanced,0.17283199230829874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,power_law_1.2,0.033983999490737916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,balanced,0.18452266852060953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,power_law_1.01,0.13252480030059816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,balanced,0.20353599389394125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,balanced,0.15042666594187418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,power_law_1.2,0.03527039885520935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,power_law_1.01,1.3700096130371093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,power_law_1.2,2.6037567138671873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,power_law_1.01,0.17310080528259278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,balanced,0.1360373298327128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,balanced,0.13526399930318198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,balanced,0.09915199875831604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,power_law_1.01,0.18564480543136597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,power_law_1.2,0.022047999501228332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,power_law_1.01,0.21101438999176025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,power_law_1.2,0.025836798548698425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,power_law_1.01,1.8633663177490234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,power_law_1.2,0.034508800506591795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,power_law_1.01,0.25734400749206543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,power_law_1.2,0.046777600049972536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,power_law_1.01,0.30792961120605467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,power_law_1.2,1.559001636505127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,power_law_1.2,0.0639743983745575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,power_law_1.01,0.3611520051956177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,power_law_1.2,0.02215680032968521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,power_law_1.01,0.08224639892578126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,power_law_1.2,0.08196480274200439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,power_law_1.01,2.482975959777832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,power_law_1.01,0.4749567985534668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,power_law_1.01,0.08235520124435425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,balanced,0.22823466857274374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,power_law_1.2,0.10575360059738159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,balanced,0.2714879910151164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,power_law_1.01,0.10175360441207885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,balanced,0.3151786724726359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,power_law_1.2,0.03832319974899292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,power_law_1.01,0.5949183940887451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,balanced,0.47575998306274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,power_law_1.01,0.10455039739608765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,power_law_1.2,0.03765760064125061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,balanced,0.5627786715825399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,power_law_1.01,0.10652799606323242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,power_law_1.01,0.8333120346069336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,power_law_1.2,0.04497919976711273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,power_law_1.01,0.11019519567489625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,power_law_1.2,0.08262400031089782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,power_law_1.01,1.0723584175109864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,power_law_1.2,0.07562239766120911
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,power_law_1.2,0.06198400259017944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,power_law_1.2,0.02391680032014847
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,power_law_1.2,0.05428479909896851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,power_law_1.2,0.02720640003681183
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,power_law_1.2,0.052300798892974856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,balanced,0.09988799691200256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,power_law_1.2,0.10848000049591064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,balanced,0.12065600355466206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,power_law_1.2,0.03406080007553101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,balanced,0.11961066722869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,power_law_1.2,0.05318400263786316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,balanced,0.1218346655368805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,power_law_1.2,0.08647680282592773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,balanced,0.12586133678754172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,balanced,0.12621333201726279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,power_law_1.2,0.043961599469184875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,power_law_1.2,0.059308797121047974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,balanced,0.1288746694723765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,power_law_1.2,0.08789759874343872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,balanced,0.13571199774742126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,balanced,0.17640000581741333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,power_law_1.2,0.05324800014495849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,power_law_1.2,0.11086080074310303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,balanced,0.18492799997329712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,power_law_1.01,0.11648000478744507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,balanced,0.2047040065129598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,balanced,0.22381333510080972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,power_law_1.01,0.11844480037689209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,power_law_1.2,0.07831040024757385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,balanced,0.2598080039024353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,power_law_1.01,0.12736639976501465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,balanced,0.297050674756368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,power_law_1.2,0.08108159899711609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,balanced,0.45338133970896405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,power_law_1.01,0.16913919448852538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,power_law_1.2,0.08204159736633301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,balanced,0.5299040079116821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,power_law_1.01,0.1835968017578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,power_law_1.2,0.08330879807472229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,balanced,0.7640799681345621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,power_law_1.01,0.21651840209960938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,power_law_1.2,0.10081280469894409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,power_law_1.01,0.2531264066696167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,power_law_1.2,0.10325759649276733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,balanced,0.8106239636739095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,power_law_1.01,0.31296000480651853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,power_law_1.2,0.06527360081672669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,balanced,1.0625813007354736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,power_law_1.01,0.37128961086273193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,power_law_1.2,0.1118399977684021
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,power_law_1.2,0.07824640274047852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,power_law_1.2,0.11358720064163208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,power_law_1.01,0.49733757972717285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,power_law_1.2,0.09388160109519958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,power_law_1.2,0.11832959651947021
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,power_law_1.2,0.11921919584274292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,power_law_1.2,0.12298239469528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,power_law_1.01,0.6205183982849121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,power_law_1.2,0.12639360427856444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,power_law_1.2,0.15081599950790406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,power_law_1.2,0.13466880321502686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,power_law_1.2,0.20449280738830566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,power_law_1.01,0.8710016250610352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,power_law_1.2,0.17612799406051635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,power_law_1.2,0.2697279930114746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,power_law_1.2,0.10590720176696777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,power_law_1.2,0.18913919925689698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,power_law_1.01,1.1228672027587892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,power_law_1.2,0.11057920455932617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,power_law_1.2,0.2337791919708252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,power_law_1.2,0.40607361793518065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,power_law_1.2,0.1158784031867981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,power_law_1.2,0.26552960872650144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,power_law_1.2,0.12066559791564942
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,power_law_1.2,0.5228735923767089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,power_law_1.2,0.3152384042739868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,power_law_1.2,0.1299072027206421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,power_law_1.2,0.17333120107650757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,balanced,1.0025813579559326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,power_law_1.2,0.18936320543289184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,power_law_1.2,0.22795519828796387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,power_law_1.2,0.2601471900939941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,32,balanced,0.03287466615438461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,32,balanced,0.03153600047032038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,32,balanced,0.0336053321758906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,32,balanced,0.04177066683769226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,32,balanced,0.041749333341916404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,32,balanced,0.05198400219281515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,32,balanced,0.05700799822807312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,32,balanced,0.05811200042565664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,32,balanced,0.0709386666615804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,32,balanced,0.07044266661008199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,32,balanced,0.07178133229414622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,32,balanced,0.0724426656961441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,power_law_1.2,0.36788480281829833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,32,balanced,0.07438933352629344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,32,balanced,0.07659199833869934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,32,balanced,0.08191466828187306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,32,balanced,0.08505066235860188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,32,balanced,0.09321600198745728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,power_law_1.2,0.48044161796569823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,32,balanced,0.11369599898656209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,32,balanced,0.12702932953834534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,32,balanced,0.15567466616630554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,power_law_1.2,0.6028351783752441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,32,balanced,0.029482667644818623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,power_law_1.2,0.31734399795532225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,32,balanced,0.03162666658560435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,power_law_1.2,0.8417856216430664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,32,balanced,0.031557333966096245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,32,balanced,0.03540800015131632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,32,balanced,0.037791999677817024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,32,balanced,0.039850667119026184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,power_law_1.2,0.3778496026992798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,32,balanced,0.04772266745567322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,32,balanced,0.047728002071380615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,32,balanced,0.05384533107280731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,32,balanced,0.05584533512592316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,power_law_1.2,1.087014389038086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,32,balanced,0.0620959997177124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,power_law_1.2,0.5032127857208252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,32,balanced,0.06614933411280315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,32,balanced,0.07008000214894612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,32,power_law_1.01,0.04547199904918671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,32,balanced,0.07594133416811626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,32,balanced,0.08070933322111766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,power_law_1.2,0.6294784069061279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,32,power_law_1.01,0.045587199926376346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,32,power_law_1.01,0.04664320051670075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,32,balanced,0.1880906621615092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,32,power_law_1.01,0.033939200639724734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,32,balanced,0.26519999901453656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,power_law_1.2,0.8801535606384278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,32,balanced,0.32265599568684894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,32,power_law_1.01,0.03473919928073883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,32,balanced,0.45334935188293457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,32,power_law_1.01,0.0343423992395401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,32,balanced,0.570741335550944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,power_law_1.2,1.1368063926696776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,32,power_law_1.01,0.03470720052719116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,32,balanced,0.8262080351511637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,32,power_law_1.2,0.045875200629234315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,32,power_law_1.01,0.03547520041465759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,32,power_law_1.2,0.0350271999835968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,32,power_law_1.2,0.045772799849510194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,32,power_law_1.01,0.03815680146217346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,32,balanced,1.0738666852315266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,32,power_law_1.01,0.04516479969024658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,32,power_law_1.2,0.04707840085029602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,32,balanced,0.0846506655216217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,32,power_law_1.01,0.04650239944458008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,32,power_law_1.2,0.04531840085983276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,32,power_law_1.01,0.05130879878997803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,32,power_law_1.2,0.04336000084877014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,32,power_law_1.01,0.05311359763145447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,32,power_law_1.2,0.049497601389884946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,32,power_law_1.01,0.06165760159492493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,32,power_law_1.2,0.05607680082321167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,32,power_law_1.01,0.06558719873428345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,32,power_law_1.2,0.05630080103874206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,32,power_law_1.01,0.06772480010986329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,32,power_law_1.2,0.0692031979560852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,32,power_law_1.01,0.07523840069770812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,32,power_law_1.2,0.06798719763755798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,32,power_law_1.01,0.08042240142822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,32,power_law_1.2,0.07006080150604248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,32,power_law_1.01,0.08835840225219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,32,power_law_1.2,0.035743999481201175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,32,power_law_1.2,0.0683135986328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,32,power_law_1.01,0.10032000541687011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,32,balanced,0.09674666325251262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,32,power_law_1.2,0.03583360016345978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,32,balanced,0.11690133810043335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,32,power_law_1.01,0.11799039840698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,32,balanced,0.13169599572817484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,32,power_law_1.2,0.0351936012506485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,32,balanced,0.1639786660671234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,32,power_law_1.01,0.13351039886474608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,32,power_law_1.01,0.04578559994697571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,32,balanced,0.19749333461125693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,32,power_law_1.2,0.036575999855995175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,32,power_law_1.01,0.17345279455184937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,32,power_law_1.01,0.046323201060295104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,32,power_law_1.2,0.039059200882911684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,32,power_law_1.01,0.04663040041923523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,32,power_law_1.01,0.21410560607910156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,32,power_law_1.2,0.045952001214027406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,32,power_law_1.01,0.05379840135574341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,32,power_law_1.2,0.04707199931144714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,32,power_law_1.01,0.296889591217041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,32,power_law_1.01,0.055776000022888184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,32,power_law_1.2,0.05321599841117859
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,32,power_law_1.01,0.37690880298614504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,32,power_law_1.2,0.05294079780578613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,32,power_law_1.2,0.06178560256958008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,32,power_law_1.01,0.5255680084228516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,32,power_law_1.2,0.07122560143470764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,32,1,balanced,0.029487999776999157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,32,1,balanced,0.03769599894682566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,32,1,balanced,0.046570668617884316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,32,power_law_1.2,0.0777728021144867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,32,power_law_1.01,0.6866432189941406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,32,power_law_1.2,0.0818943977355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,32,balanced,0.2823359966278076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,32,power_law_1.2,0.08610560297966004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,32,balanced,0.35072000821431476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,32,power_law_1.01,0.9950976371765137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,32,power_law_1.2,0.10084480047225952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,32,balanced,0.49143465360005695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,32,power_law_1.2,0.13045120239257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,32,balanced,0.6246986786524454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,32,power_law_1.2,0.14394240379333495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,32,power_law_1.01,1.2977472305297852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,32,power_law_1.01,0.06840320229530335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,32,1,power_law_1.01,0.027315199375152588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,32,power_law_1.2,0.1887295961380005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,32,power_law_1.01,0.06897280216217042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,32,1,power_law_1.01,0.036236798763275145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,32,power_law_1.01,0.06803200244903565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,32,power_law_1.2,0.23209600448608397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,32,power_law_1.2,0.06647040247917176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,32,1,power_law_1.01,0.04751999974250794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,32,power_law_1.01,0.07026559710502625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,32,power_law_1.2,0.309932804107666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,32,power_law_1.2,0.06815999746322632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,32,power_law_1.01,0.07077760100364686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,32,power_law_1.2,0.07735040187835693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,32,power_law_1.01,0.07616639733314515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,32,power_law_1.2,0.08368639945983887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,32,power_law_1.01,0.08107519745826722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,32,power_law_1.2,0.08941439986228943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,32,power_law_1.01,0.0855679988861084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,32,power_law_1.2,0.10088319778442383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,32,power_law_1.01,0.09709439873695373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,32,power_law_1.2,0.12372479438781739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,32,balanced,0.9058026472727457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,32,power_law_1.2,0.14300800561904908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,32,power_law_1.2,0.18165119886398315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,32,balanced,1.1804746786753337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,32,power_law_1.2,0.22796800136566162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,32,1,balanced,0.08693333466847737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,32,1,balanced,0.1563093364238739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,32,1,balanced,0.029535998900731403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,32,power_law_1.2,0.31614720821380615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,32,1,balanced,0.20593599478403726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,32,1,balanced,0.03965866565704346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,32,1,balanced,0.05601066847642263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,32,power_law_1.2,0.40067200660705565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,32,power_law_1.2,0.5599040031433106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,32,power_law_1.01,0.11864320039749146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,32,power_law_1.01,0.14053759574890137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,32,power_law_1.2,0.7215680122375489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,32,power_law_1.01,0.17719680070877075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,32,power_law_1.01,0.21502079963684081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,32,1,power_law_1.01,0.07880319952964783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,32,power_law_1.01,0.29539198875427247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,32,power_law_1.2,0.4002175807952881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,32,1,power_law_1.01,0.10164480209350586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,32,power_law_1.01,0.3625152111053467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,32,1,power_law_1.01,0.16714240312576295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,32,power_law_1.2,0.5704448223114014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,32,power_law_1.01,0.5267263889312744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,32,power_law_1.01,0.6776895999908448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,32,power_law_1.01,0.9580927848815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,32,power_law_1.01,1.3416704177856444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,32,1,balanced,0.0869653324286143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,32,1,balanced,0.15061333775520325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,32,1,power_law_1.2,0.027846398949623107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,32,1,balanced,0.19338667392730713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,32,1,power_law_1.2,0.03743999898433685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,32,1,balanced,0.22291199366251627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,32,1,balanced,0.22387200593948364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,32,1,power_law_1.2,0.05489919781684875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,32,1,power_law_1.2,0.07374719977378845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,32,1,power_law_1.2,0.09253759980201721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,32,1,power_law_1.2,0.1504256010055542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,32,1,power_law_1.2,0.18316160440444945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,32,1,power_law_1.2,0.18716800212860107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,32,1,power_law_1.2,0.6725056171417236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,32,1,power_law_1.2,0.6925504207611084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,32,1,power_law_1.01,0.1814463973045349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,32,1,power_law_1.2,0.7994112014770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,32,1,power_law_1.01,0.1926911950111389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,32,1,power_law_1.2,0.2581568002700806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,32,1,power_law_1.2,0.26829440593719484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,32,1,power_law_1.2,0.2822335958480835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,32,1,power_law_1.01,0.027775999903678895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,32,1,power_law_1.2,0.2994175910949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,32,1,power_law_1.01,0.03734399974346161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,32,1,power_law_1.2,0.31261439323425294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,32,1,power_law_1.01,0.05511680245399475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,32,1,power_law_1.2,0.3465536117553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,32,power_law_1.2,0.7555456161499023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,32,1,power_law_1.2,0.7235583782196044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,32,power_law_1.2,1.112934398651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,32,1,power_law_1.2,0.7867136001586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,32,1,balanced,0.19524800777435303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,32,1,balanced,0.19534399112065634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,32,1,power_law_1.2,0.9143808364868165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,32,power_law_1.2,1.469644832611084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,32,1,power_law_1.2,1.033529567718506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,32,power_law_1.2,1.080735969543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,32,1,balanced,0.21604265769322714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,32,1,power_law_1.2,0.026131200790405273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,32,1,power_law_1.2,1.263923168182373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,32,1,balanced,0.21809599796930948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,32,1,power_law_1.2,0.036345601081848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,32,1,power_law_1.2,0.04709759950637817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,32,1,power_law_1.2,1.4950271606445313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,32,1,power_law_1.01,0.1655295968055725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,32,1,power_law_1.2,1.9784511566162108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,32,1,power_law_1.01,0.16922240257263182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,32,balanced,0.03985599925120672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,32,balanced,0.06452266871929169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,32,balanced,0.08572266499201457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,32,balanced,0.15461333592732748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,32,balanced,0.22403732935587564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,32,balanced,0.25066133340199787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,32,balanced,0.24236800273259482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,32,balanced,0.24861333767573038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,32,balanced,0.254805326461792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,32,balanced,0.25194666783014935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,32,balanced,0.39295466740926105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,32,balanced,0.41359468301137287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,32,balanced,0.4275786479314168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,32,balanced,0.44969598452250165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,32,balanced,0.4696799914042155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,32,balanced,0.4928106864293416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,32,balanced,0.5385973453521729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,32,balanced,0.6139786640803019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,32,balanced,0.6853386561075846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,32,balanced,0.8274986743927002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,32,balanced,0.9679146607716879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,32,1,balanced,0.7483839988708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,32,1,balanced,0.7482826709747314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,32,1,balanced,0.1898826758066813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,32,power_law_1.2,1.3853440284729004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,32,balanced,0.03737066686153412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,32,balanced,0.05205333232879639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,32,balanced,0.06406933565934499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,32,balanced,0.09855467081069946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,32,balanced,0.14666133125623068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,32,balanced,0.15440000096956888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,32,balanced,0.17402666807174683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,32,balanced,0.1787573297818502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,32,balanced,0.1653279960155487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,32,1,power_law_1.2,2.449433517456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,32,balanced,0.1701386570930481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,32,balanced,0.33133333921432495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,32,balanced,0.350383996963501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,32,1,power_law_1.01,0.0782912015914917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,32,balanced,0.3674079974492391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,32,balanced,0.3906240065892537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,32,1,power_law_1.01,0.09891200065612793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,32,balanced,0.4126400152842204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,32,1,power_law_1.01,0.15204479694366455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,32,balanced,0.4395039876302083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,32,balanced,0.4878079891204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,32,1,power_law_1.2,3.410598373413086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,32,1,power_law_1.01,0.16606719493865968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,32,balanced,0.5584959983825684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,32,balanced,1.3544425964355469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,32,1,power_law_1.2,0.076665598154068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,32,balanced,0.6409973303476969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,32,1,power_law_1.2,0.09598720073699951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,32,balanced,0.7999306519826254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,32,balanced,1.6398720741271973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,32,1,power_law_1.2,0.16047359704971315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,32,balanced,0.9596906503041586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,32,balanced,1.3323252995808919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,32,1,balanced,0.1849600076675415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,32,balanced,2.324069341023763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,32,1,balanced,0.18736000855763754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,32,1,power_law_1.2,4.385305786132813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,32,1,balanced,0.18838934103647867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,32,1,balanced,0.19122666120529175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,32,1,balanced,0.193615992863973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,32,balanced,1.649295965830485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,32,1,balanced,0.19946134090423584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,32,balanced,0.05208533505598704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,32,balanced,0.07018666466077168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,32,balanced,0.07635200023651123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,32,balanced,0.10470400253931682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,32,balanced,0.1458613375822703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,32,balanced,0.15634666879971823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,32,balanced,0.16446399688720703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,32,balanced,2.344634691874186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,32,balanced,0.17045867443084717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,32,balanced,0.3823253313700358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,32,balanced,0.39048532644907635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,32,balanced,0.407696008682251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,32,balanced,0.42551998297373456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,32,balanced,2.9804906845092773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,32,balanced,0.44438934326171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,32,balanced,0.47411199410756427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,32,1,balanced,0.686410665512085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,32,balanced,0.5039253234863281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,32,balanced,0.5386399825414022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,32,balanced,2.904613176981608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,32,1,power_law_1.01,0.1709247946739197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,32,1,power_law_1.01,0.17294080257415773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,32,balanced,4.414592107137044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,32,1,power_law_1.01,0.17932800054550171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,32,1,power_law_1.01,0.18734079599380493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,32,1,power_law_1.01,0.19673600196838378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,32,1,power_law_1.01,0.20861439704895018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,32,balanced,4.319610595703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,32,balanced,0.6019466718037924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,32,balanced,5.754474639892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,32,balanced,0.7227893670399984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,32,balanced,0.8484906355539957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,32,power_law_1.01,0.19920639991760253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,32,balanced,5.585072199503581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,32,balanced,1.2056159973144531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,32,power_law_1.01,0.1985152006149292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,32,power_law_1.01,0.13147519826889037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,32,balanced,1.4506239891052246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,32,power_law_1.01,0.19955840110778808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,32,power_law_1.01,0.13148800134658814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,32,balanced,2.0548906326293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,32,power_law_1.01,0.20349440574645997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,32,power_law_1.01,0.13254400491714477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,32,power_law_1.01,0.21928319931030274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,32,balanced,2.546010653177897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,32,power_law_1.01,0.13640960454940795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,32,power_law_1.01,0.23466238975524903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,32,power_law_1.01,0.14257279634475709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,32,power_law_1.01,0.23097600936889648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,32,1,balanced,0.21638933817545572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,32,1,balanced,0.223962664604187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,32,power_law_1.01,0.1499392032623291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,32,power_law_1.01,0.21747839450836182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,32,balanced,3.7497278849283853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,32,1,balanced,0.2794400056203206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,32,power_law_1.01,0.2114880084991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,32,1,balanced,0.28273600339889526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,32,1,balanced,0.2898133397102356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,32,1,balanced,0.29543999830881756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,32,power_law_1.01,0.23071999549865724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,32,1,balanced,0.2998933394749959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,32,1,balanced,0.30829334259033203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,32,power_law_1.01,0.38321919441223146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,32,power_law_1.01,0.4011839866638184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,32,balanced,4.841861406962077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,32,power_law_1.01,0.41211519241333006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,32,1,balanced,0.24148799975713095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,32,power_law_1.01,0.43559679985046384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,32,1,balanced,0.2588160037994385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,32,power_law_1.01,0.16176639795303344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,32,1,balanced,0.4382559855779012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,32,power_law_1.01,0.4606656074523926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,32,1,balanced,0.4682933489481608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,32,power_law_1.01,0.16469759941101075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,32,1,balanced,0.6787839730580648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,32,power_law_1.01,0.4823552131652832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,32,power_law_1.01,0.1487231969833374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,32,1,balanced,0.746234655380249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,32,power_law_1.01,0.5480639934539795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,32,power_law_1.01,0.16906239986419677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,32,1,balanced,1.1693066755930583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,32,1,power_law_1.01,0.182150399684906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,32,1,power_law_1.01,0.2224384069442749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,32,power_law_1.01,0.6330111980438232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,32,1,power_law_1.01,0.1823807954788208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,32,1,power_law_1.01,0.2437376022338867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,32,1,balanced,1.4586505889892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,32,1,power_law_1.01,0.3024319887161255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,32,power_law_1.01,0.7249728202819824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,32,1,power_law_1.2,0.1792448043823242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,32,1,power_law_1.01,0.3405632019042969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,32,1,power_law_1.2,0.18780159950256348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,32,1,power_law_1.01,0.4215360164642334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,32,power_law_1.01,0.8953215599060058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,32,1,power_law_1.01,0.49894399642944337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,32,power_law_1.01,1.089139175415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,32,1,power_law_1.01,0.6781248092651367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,32,balanced,7.171578725179036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,32,power_law_1.01,0.32718079090118407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,32,1,power_law_1.01,0.8558015823364258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,32,power_law_1.01,1.4488256454467774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,32,power_law_1.01,0.3482111930847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,32,power_law_1.01,0.3650559902191162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,32,1,power_law_1.01,1.2012415885925294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,32,power_law_1.01,0.38760321140289306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,32,power_law_1.01,1.7983104705810546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,32,power_law_1.01,0.41173758506774905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,32,1,power_law_1.01,1.550291156768799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,32,power_law_1.01,0.43885440826416017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,32,power_law_1.01,0.4893951892852783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,32,balanced,9.365045547485352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,32,power_law_1.01,2.531488037109375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,32,power_law_1.01,0.5718783855438232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,32,power_law_1.2,0.1320512056350708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,32,power_law_1.01,0.6591807842254639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,32,power_law_1.2,0.1318336009979248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,32,power_law_1.01,0.8434687614440918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,32,power_law_1.2,0.13306879997253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,32,power_law_1.01,3.188710403442383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,32,power_law_1.01,1.0145024299621581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,32,1,balanced,0.6744159857432047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,32,power_law_1.01,4.836991882324218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,32,1,balanced,0.6833386421203613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,32,power_law_1.2,0.1982975959777832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,32,1,balanced,0.7008693218231201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,32,power_law_1.2,0.1359935998916626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,32,1,balanced,0.71451735496521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,32,power_law_1.2,0.19713280200958253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,32,power_law_1.2,0.1428607940673828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,32,1,balanced,1.323093334833781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,32,power_law_1.2,0.19913599491119385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,32,power_law_1.2,0.1397503972053528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,32,1,balanced,1.3541547457377117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,32,power_law_1.2,0.20245759487152098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,32,power_law_1.2,0.16357120275497436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,32,1,balanced,1.992634614308675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,32,power_law_1.2,0.1720576047897339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,32,power_law_1.2,0.2089855909347534
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,32,power_law_1.2,0.14514559507369995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,32,power_law_1.01,6.326508712768555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,32,power_law_1.01,0.1368127942085266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,32,1,balanced,2.0557920138041177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,32,power_law_1.2,0.23899519443511963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,32,power_law_1.2,0.17144960165023804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,32,power_law_1.2,0.2303231954574585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,32,power_law_1.01,0.13747199773788452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,32,power_law_1.2,0.13923200368881225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,32,power_law_1.2,0.23417599201202394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,32,power_law_1.01,0.1392832040786743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,32,power_law_1.2,0.1376703977584839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,32,power_law_1.2,0.2241152048110962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,32,power_law_1.2,0.22539520263671875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,32,power_law_1.2,0.38277759552001955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,32,power_law_1.2,0.39779839515686033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,32,power_law_1.2,0.3265023946762085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,32,power_law_1.01,1.375494384765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,32,power_law_1.2,0.4130112171173096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,32,power_law_1.2,0.34851200580596925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,32,1,balanced,3.3296639124552407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,32,power_law_1.2,0.4430079936981201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,32,power_law_1.2,0.36253440380096436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,32,power_law_1.01,1.733964729309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,32,power_law_1.2,0.4616960048675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,32,power_law_1.2,0.13848960399627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,32,power_law_1.2,0.3874624013900757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,32,power_law_1.2,0.4955776214599609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,32,power_law_1.2,0.41259517669677737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,32,1,balanced,4.036266644795735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,32,power_law_1.2,0.1395967960357666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,32,power_law_1.2,0.5459072113037109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,32,power_law_1.2,0.44309120178222655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,32,power_law_1.2,0.1351807951927185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,32,power_law_1.01,2.4721471786499025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,32,power_law_1.2,0.6510144233703613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,32,power_law_1.2,0.5003712177276611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,32,1,power_law_1.2,0.1637760043144226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,32,1,power_law_1.2,0.17704319953918457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,32,power_law_1.2,0.7368832111358643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,32,power_law_1.2,0.5698048114776612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,32,power_law_1.2,0.667091178894043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,32,power_law_1.2,0.9332096099853515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,32,power_law_1.2,0.8589632034301757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,32,power_law_1.01,0.1411967992782593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,32,power_law_1.2,1.130342388153076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,32,power_law_1.2,0.1327679991722107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,32,power_law_1.2,1.0507904052734376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,32,power_law_1.01,0.14502400159835815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,32,power_law_1.2,0.14588799476623535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,32,power_law_1.2,0.15319039821624755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,32,power_law_1.01,3.1785024642944335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,32,power_law_1.2,0.3736191987991333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,32,1,balanced,0.061306665341059365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,32,1,balanced,0.09920000036557515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,32,1,balanced,0.18387732903162637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,32,power_law_1.2,0.37897601127624514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,32,power_law_1.2,0.3950655937194824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,32,power_law_1.2,0.40879359245300295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,32,power_law_1.2,0.4359488010406494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,32,power_law_1.2,0.4743807792663574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,32,power_law_1.01,4.652083206176758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,32,power_law_1.2,0.514412784576416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,32,power_law_1.2,0.5622848033905029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,32,power_law_1.2,0.6447296142578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,32,power_law_1.2,0.8231871604919434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,32,power_law_1.01,6.078694534301758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,32,power_law_1.2,0.9764991760253906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,32,1,balanced,0.42899731794993085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,32,power_law_1.01,0.14245760440826416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,32,power_law_1.2,1.4942272186279297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,32,1,balanced,0.8124266465504965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,32,power_law_1.2,1.3087039947509767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,32,1,balanced,1.1791573365529378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,32,power_law_1.2,1.9803647994995117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,32,power_law_1.2,1.6831935882568358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,32,power_law_1.2,1.4164928436279296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,32,power_law_1.2,2.4227840423583986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,32,power_law_1.2,2.6725759506225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,32,power_law_1.2,1.7832128524780273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,32,power_law_1.01,0.1493824005126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,32,power_law_1.01,0.15105279684066772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,32,1,power_law_1.01,0.6565887928009033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,32,power_law_1.2,3.0338048934936523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,32,power_law_1.2,2.532364845275879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,32,1,balanced,0.06876266499360402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,32,1,balanced,0.11378133296966553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,32,1,power_law_1.01,0.676416015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,32,1,balanced,0.20964799324671426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,32,power_law_1.2,3.3750720977783204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,32,power_law_1.2,4.4820606231689455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,32,1,balanced,1.2649493217468262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,32,power_law_1.2,3.5162559509277345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,32,1,balanced,1.2680426438649495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,32,power_law_1.2,4.843040084838867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,32,power_law_1.2,5.897881698608399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,32,power_law_1.2,5.227097702026367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,32,1,balanced,0.4075466791788737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,32,1,balanced,0.7759520212809244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,32,1,balanced,1.1374239921569824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,32,power_law_1.2,6.447468566894531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,32,power_law_1.2,8.872358703613282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,32,1,balanced,1.215114672978719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,32,1,balanced,1.215567986170451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,32,power_law_1.2,6.697510528564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,32,power_law_1.01,0.373689603805542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,32,power_law_1.01,0.37946879863739014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,32,power_law_1.01,0.4016704082489014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,32,1,balanced,1.2570772965749104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,32,power_law_1.01,0.4161215782165527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,32,1,balanced,1.2649919986724854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,32,power_law_1.01,0.4374847888946533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,32,1,power_law_1.01,0.05902720093727112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,32,power_law_1.01,0.46460800170898436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,32,1,power_law_1.01,0.0964031994342804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,32,power_law_1.01,0.5078080177307129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,32,1,power_law_1.01,0.18288639783859253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,32,power_law_1.01,0.5413440227508545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,32,power_law_1.01,0.6382783889770508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,32,power_law_1.01,0.7923456192016601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,32,power_law_1.01,0.9555199623107911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,32,1,power_law_1.01,0.0686016023159027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,32,1,power_law_1.01,0.11201280355453491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,32,power_law_1.01,1.2825920104980468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,32,1,power_law_1.01,0.21118719577789308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,32,power_law_1.2,11.636544036865235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,32,power_law_1.01,1.5809727668762208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,32,1,power_law_1.2,0.1750208020210266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,32,power_law_1.01,2.3093120574951174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,32,1,power_law_1.01,0.38537600040435793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,32,1,power_law_1.01,0.5339263916015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,32,1,balanced,1.306165297826131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,32,power_law_1.01,2.96694393157959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,32,1,power_law_1.01,0.8814592361450195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,32,1,balanced,4.521327972412109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,32,power_law_1.01,4.157548904418945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,32,1,power_law_1.2,0.05940480232238769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,32,1,power_law_1.01,0.3628927946090698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,32,1,balanced,4.51910940806071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,32,1,power_law_1.2,0.1019711971282959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,32,1,power_law_1.01,0.5120255947113037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,32,1,power_law_1.2,0.1832383990287781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,32,1,power_law_1.01,0.8661375999450683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,32,power_law_1.01,5.496518325805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,32,1,power_law_1.01,1.0163519859313965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,32,power_law_1.01,8.135199737548827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,32,1,power_law_1.2,0.3319231986999512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,32,1,power_law_1.2,0.48874239921569823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,32,1,power_law_1.01,1.0147647857666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,32,1,power_law_1.2,0.8717951774597168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,32,1,balanced,1.1388320128122966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,32,1,power_law_1.01,1.055692768096924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,32,1,balanced,1.1499679883321126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,32,1,balanced,1.159333308537801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,32,1,balanced,1.1690719922383626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,32,1,balanced,1.1833439668019612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,32,1,balanced,1.2051040331522624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,32,1,power_law_1.01,1.0286784172058105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,32,power_law_1.01,10.736627197265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,32,1,balanced,5.200517336527507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,32,1,power_law_1.2,0.17406079769134522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,32,1,power_law_1.2,0.17787519693374634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,32,1,power_law_1.2,0.18050559759140014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,32,1,power_law_1.2,0.19219839572906494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,32,1,power_law_1.2,0.2
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,32,1,power_law_1.2,0.21671679019927978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,32,1,power_law_1.2,1.005241584777832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,32,1,power_law_1.2,1.052780818939209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,32,1,power_law_1.01,1.0266559600830079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,32,1,power_law_1.01,1.0491135597229004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,32,1,power_law_1.2,0.06796799898147583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,32,1,power_law_1.2,0.11582080125808716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,32,1,power_law_1.2,0.21035521030426024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,32,1,balanced,1.768496036529541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,32,1,balanced,1.3990720113118489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,32,1,power_law_1.01,3.7717121124267576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,32,1,balanced,1.7805066108703613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,32,1,balanced,1.432096004486084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,32,1,balanced,1.4977653821309407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,32,1,balanced,1.7882453600565593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,32,1,power_law_1.2,1.0339648246765136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,32,1,balanced,1.5703253746032715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,32,1,balanced,1.8008480072021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,32,1,power_law_1.01,3.9055999755859374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,32,1,power_law_1.2,1.058732795715332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,32,1,balanced,1.8105120658874512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,32,1,balanced,2.817882537841797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,32,1,power_law_1.2,0.33741440773010256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,32,1,balanced,1.8357280095418294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,32,1,power_law_1.2,0.46390399932861326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,32,1,balanced,2.9557065963745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,32,1,power_law_1.2,0.8622591972351075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,32,1,balanced,4.3548533121744795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,32,1,balanced,4.673904101053874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,32,1,power_law_1.01,1.1629695892333984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,32,1,balanced,7.555967966715495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,32,1,power_law_1.2,0.9965056419372559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,32,1,power_law_1.2,1.0388928413391114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,32,1,balanced,9.193056106567383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,64,balanced,0.053616002202034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,64,balanced,0.05201066533724467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,64,balanced,0.052005335688591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,64,balanced,0.05189333359400431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,64,balanced,0.07241599758466084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,64,balanced,0.11131200194358826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,64,balanced,0.11843732992808025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,64,balanced,0.12109866738319397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,64,balanced,0.1307360033194224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,64,balanced,0.13276267051696777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,64,balanced,0.18092799186706543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,64,balanced,0.18905067443847656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,64,balanced,0.1938719948132833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,64,balanced,0.20253866910934448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,64,balanced,0.21157866716384888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,64,balanced,0.2193653384844462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,64,balanced,0.23675199349721274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,32,1,balanced,4.007482528686523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,64,balanced,0.28126933177312213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,64,balanced,0.3122239907582601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,32,1,power_law_1.01,1.0248191833496094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,64,balanced,0.3721333344777425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,64,balanced,0.43401598930358887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,32,1,power_law_1.01,4.50398063659668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,32,1,power_law_1.2,1.1798912048339845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,64,balanced,0.5556373198827108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,32,1,power_law_1.01,1.0519167900085449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,32,1,balanced,4.03221321105957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,64,balanced,0.6787306467692057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,64,balanced,0.969754695892334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,32,1,power_law_1.01,1.087052822113037
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,32,1,power_law_1.2,0.23118081092834472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,64,balanced,1.214128017425537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,32,1,power_law_1.2,0.27305600643157957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,32,1,power_law_1.01,1.106719970703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,32,1,balanced,4.107237180074056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,32,1,power_law_1.2,0.3075200080871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,64,balanced,1.7425920168558757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,32,1,power_law_1.2,0.3506367921829224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,32,1,power_law_1.01,1.1611583709716797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,32,1,power_law_1.2,0.42552962303161623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,64,balanced,2.2792533238728843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,32,1,power_law_1.01,1.2506239891052247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,64,balanced,0.04162666698296865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,64,balanced,0.04161066561937332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,64,balanced,0.04254400233427683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,64,balanced,0.0418453315893809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,64,balanced,0.05416533350944519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,32,1,power_law_1.2,0.5042687892913819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,32,1,power_law_1.01,0.7691199779510498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,32,1,power_law_1.2,0.6884543895721436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,32,1,power_law_1.2,0.8718912124633789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,32,1,balanced,4.188991864522298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,32,1,power_law_1.2,3.7234622955322267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,64,balanced,0.0809333324432373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,64,balanced,0.09810133775075276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,64,balanced,0.10037866234779358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,64,balanced,0.10173867146174113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,64,balanced,0.10517866412798564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,64,balanced,0.16262933611869812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,64,balanced,0.1728853384653727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,64,balanced,0.17836799224217734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,64,balanced,0.18703999121983847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,64,balanced,0.19966934124628702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,64,balanced,0.2097919980684916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,32,1,power_law_1.2,3.8505855560302735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,64,balanced,0.23237866163253784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,64,balanced,0.27109332879384357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,64,balanced,0.30565333366394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,64,balanced,0.3727733294169108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,32,1,balanced,7.970437367757161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,64,balanced,0.4410239855448405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,64,balanced,0.5754400094350179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,32,1,power_law_1.01,1.5828607559204102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,64,balanced,0.711178700129191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,64,balanced,1.004805326461792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,32,1,power_law_1.01,1.6253696441650392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,64,balanced,1.2758399645487468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,32,1,power_law_1.2,1.2137344360351563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,32,1,power_law_1.2,1.0641983985900878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,32,1,power_law_1.01,1.6406911849975585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,64,balanced,1.8346667289733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,32,1,power_law_1.2,1.564121627807617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,64,balanced,0.06247466802597046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,64,balanced,0.06229866544405619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,32,1,power_law_1.2,1.0820032119750977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,64,balanced,0.06404266754786174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,64,balanced,0.06414400041103363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,64,balanced,0.07643199960390727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,32,1,power_law_1.01,1.7119871139526368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,64,balanced,2.395573298136393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,64,balanced,0.10264533758163452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,64,balanced,0.10431999961535136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,64,balanced,0.10540266831715901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,32,1,power_law_1.2,1.1361663818359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,64,balanced,0.19276267290115356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,64,balanced,0.19782400131225586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,64,balanced,0.20337599515914917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,64,balanced,0.20907733837763467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,64,balanced,0.21663999557495117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,32,1,power_law_1.2,1.1813440322875977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,32,1,balanced,8.090885162353516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,32,1,power_law_1.01,1.4605376243591308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,32,1,power_law_1.2,1.2162495613098145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,32,1,power_law_1.2,1.310918426513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,32,1,power_law_1.01,1.5652928352355957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,32,1,power_law_1.01,1.8031360626220703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,64,balanced,0.22687999407450357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,32,1,power_law_1.01,1.815020751953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,32,1,power_law_1.01,1.900761604309082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,32,1,power_law_1.01,2.0562047958374023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,32,1,balanced,12.009269714355469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,32,1,power_law_1.01,2.5703296661376953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,64,balanced,0.23884799083073935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,64,balanced,0.25150932868321735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,64,balanced,0.2798080046971639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,64,balanced,0.3289066751797994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,64,balanced,0.029493334392706554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,64,balanced,0.03156800071398417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,64,balanced,0.03126933425664902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,64,balanced,0.031583999594052635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,64,balanced,0.03350399931271871
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,64,balanced,0.04383466641108195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,64,balanced,0.06564266482988994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,32,1,power_law_1.01,3.06495361328125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,64,balanced,0.05233600238958994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,64,balanced,0.1297599971294403
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,64,balanced,0.1339413324991862
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,64,balanced,0.12239999572436015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,64,balanced,0.10847466190656026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,64,balanced,0.0939520001411438
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,64,balanced,0.07624533275763194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,64,balanced,0.06633600095907848
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,64,balanced,0.07049066821734111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,64,balanced,0.07669866581757863
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,64,balanced,0.08435199658075969
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,64,balanced,0.09294399619102478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,64,balanced,0.3800693353017171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,32,1,power_law_1.01,4.086182403564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,64,balanced,0.11310933033625285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,64,balanced,0.13191466530164084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,64,balanced,0.47868800163269043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,64,balanced,0.17292799552281699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,64,balanced,0.5806719859441122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,64,balanced,0.21548267205556235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,64,balanced,0.30668266614278156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,64,balanced,0.8194773197174072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,32,1,power_law_1.2,3.2337791442871096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,64,balanced,0.39081064860026044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,64,balanced,1.0201493104298909
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,64,balanced,0.5537013212839762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,64,balanced,0.7228480180104574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,32,1,balanced,12.31374994913737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,64,balanced,1.4575200080871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,32,1,power_law_1.01,5.114860916137696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,64,power_law_1.01,0.08798720240592957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,64,power_law_1.01,0.08796799778938294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,64,balanced,1.8922346433003743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,64,power_law_1.01,0.08835840225219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,64,power_law_1.01,0.08945279717445373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,32,1,power_law_1.2,1.5087488174438477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,64,balanced,2.76091734568278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,64,power_law_1.01,0.08210560083389282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,64,power_law_1.01,0.09695360064506531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,32,1,power_law_1.2,1.6236799240112305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,64,power_law_1.01,0.10623999834060668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,64,balanced,3.6334880193074546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,64,power_law_1.01,0.11785600185394288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,32,1,power_law_1.2,1.87640323638916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,64,power_law_1.01,0.06761599779129028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,64,power_law_1.01,0.12789119482040406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,32,1,power_law_1.01,7.235513305664062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,64,power_law_1.01,0.1294592022895813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,64,power_law_1.01,0.0675711989402771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,64,power_law_1.01,0.1791167974472046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,64,power_law_1.01,0.06641280055046081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,64,power_law_1.01,0.18392959833145142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,32,1,power_law_1.2,2.131135940551758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,64,power_law_1.01,0.06794880032539367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,32,1,power_law_1.01,4.150099182128907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,64,power_law_1.01,0.1923200011253357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,64,power_law_1.01,0.0645695984363556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,64,power_law_1.01,0.19569920301437377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,64,power_law_1.01,0.07381119728088378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,64,power_law_1.01,0.2064448118209839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,64,power_law_1.01,0.09500160217285156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,32,1,power_law_1.2,2.6371648788452147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,64,power_law_1.01,0.2162303924560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,64,power_law_1.01,0.09642239809036254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,32,1,balanced,19.951066335042317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,64,power_law_1.01,0.24038400650024414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,64,power_law_1.01,0.09566079974174499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,32,1,power_law_1.01,4.427135848999024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,64,power_law_1.01,0.1019327998161316
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,64,power_law_1.01,0.2890752077102661
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,64,power_law_1.01,0.15994240045547486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,64,power_law_1.01,0.32670719623565675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,32,1,power_law_1.2,3.1234560012817383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,64,power_law_1.01,0.1712000012397766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,32,1,power_law_1.01,9.33609619140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,64,power_law_1.01,0.4149184226989746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,32,1,power_law_1.2,1.6199296951293944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,64,power_law_1.01,0.17663999795913696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,64,power_law_1.01,0.08081279993057251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,64,power_law_1.01,0.18495999574661254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,64,power_law_1.01,0.08489599823951721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,64,power_law_1.01,0.1967039942741394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,32,1,power_law_1.2,1.635763168334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,32,1,power_law_1.01,5.020537567138672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,64,power_law_1.01,0.08541439771652222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,64,power_law_1.01,0.211411190032959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,32,1,power_law_1.2,4.154822540283203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,64,power_law_1.01,0.0853056013584137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,64,power_law_1.01,0.24015359878540038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,32,1,power_law_1.2,1.7377023696899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,64,power_law_1.01,0.084307199716568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,64,power_law_1.01,0.274182391166687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,32,1,power_law_1.01,0.2491391897201538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,64,power_law_1.01,0.09821439981460571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,64,power_law_1.01,0.3148864030838013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,32,1,power_law_1.01,0.2640127897262573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,64,power_law_1.01,0.1015936017036438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,64,power_law_1.01,0.3913088083267212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,32,1,power_law_1.2,1.7936704635620118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,64,power_law_1.01,0.5001535892486573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,64,power_law_1.01,0.10154880285263061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,64,power_law_1.01,0.4746367931365967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,32,1,power_law_1.01,5.682681655883789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,64,power_law_1.01,0.1912832021713257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,64,power_law_1.01,0.6485504150390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,64,power_law_1.01,0.19434880018234252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,32,1,power_law_1.2,5.1824897766113285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,32,1,power_law_1.2,1.8645248413085938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,64,power_law_1.01,0.8165696144104004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,64,power_law_1.01,0.2007296085357666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,64,power_law_1.01,0.20476160049438477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,64,power_law_1.01,1.1307711601257324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,64,power_law_1.01,0.2129472017288208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,32,1,power_law_1.2,1.9893823623657227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,64,power_law_1.01,0.22332799434661865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,32,1,power_law_1.01,0.27445759773254397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,64,power_law_1.01,0.23763840198516845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,32,1,power_law_1.01,0.2907520055770874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,32,1,balanced,24.082667032877605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,64,power_law_1.01,0.2552704095840454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,32,1,power_law_1.01,0.29977600574493407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,64,power_law_1.01,0.6267712116241455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,32,1,power_law_1.01,7.069126129150391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,64,power_law_1.01,0.7883391857147217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,32,1,power_law_1.2,7.319936370849609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,64,power_law_1.01,1.084819221496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,64,power_law_1.01,1.4677311897277832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,64,power_law_1.01,1.386355209350586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,64,power_law_1.01,0.2934015989303589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,32,1,power_law_1.01,0.3341952085494995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,64,power_law_1.01,0.36903679370880127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,64,power_law_1.01,2.2289920806884767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,64,power_law_1.01,0.42502398490905763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,64,power_law_1.01,2.0162111282348634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,32,1,power_law_1.01,8.334086608886718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,64,power_law_1.01,0.5661568164825439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,64,power_law_1.01,0.7101119995117188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,64,power_law_1.01,2.788159942626953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,64,power_law_1.01,2.6794559478759767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,64,power_law_1.2,0.08747519850730896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,64,power_law_1.01,0.9667584419250488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,32,1,power_law_1.2,9.43351058959961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,64,power_law_1.2,0.06675840020179749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,64,power_law_1.2,0.08763520121574402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,64,power_law_1.2,0.08105599880218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,64,power_law_1.2,0.06645119786262513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,64,power_law_1.2,0.08913919925689698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,64,power_law_1.01,1.2240896224975586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,64,power_law_1.2,0.08522239923477173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,64,power_law_1.2,0.06689280271530151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,64,power_law_1.2,0.08147839903831482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,64,power_law_1.2,0.08481280207633972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,64,power_law_1.2,0.06780160069465638
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,64,power_law_1.2,0.08213120102882385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,64,power_law_1.01,0.037350401282310486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,64,power_law_1.01,1.8585216522216796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,64,power_law_1.2,0.08529279828071594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,64,power_law_1.2,0.06216319799423218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,64,power_law_1.2,0.10032000541687011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,64,power_law_1.01,0.036959999799728395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,64,power_law_1.2,0.08162559866905213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,64,power_law_1.2,0.07660160064697266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,64,power_law_1.2,0.10609920024871826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,64,power_law_1.01,0.03678080141544342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,64,power_law_1.2,0.10142079591751099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,64,power_law_1.2,0.09661440253257751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,64,power_law_1.2,0.11274240016937256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,64,power_law_1.01,0.03866879940032959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,64,power_law_1.2,0.10236159563064576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,64,power_law_1.2,0.09589120149612426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,32,1,power_law_1.01,10.975154876708984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,64,power_law_1.2,0.1235584020614624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,64,power_law_1.01,0.036294400691986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,64,power_law_1.2,0.10129280090332031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,64,power_law_1.2,0.12688640356063843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,64,power_law_1.01,0.041126400232315063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,64,power_law_1.2,0.1915711998939514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,64,power_law_1.2,0.1768064022064209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,64,power_law_1.2,0.19486720561981202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,64,power_law_1.2,0.18587520122528076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,64,power_law_1.2,0.20165760517120362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,64,power_law_1.2,0.19173120260238646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,64,power_law_1.2,0.206496000289917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,64,power_law_1.2,0.2128000020980835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,64,power_law_1.01,2.375308799743652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,64,power_law_1.2,0.2237312078475952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,64,power_law_1.2,0.09624320268630981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,32,1,power_law_1.2,4.272921752929688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,64,power_law_1.2,0.2410304069519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,64,power_law_1.2,0.09904639720916748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,64,power_law_1.2,0.2663167953491211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,64,power_law_1.2,0.15915520191192628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,64,power_law_1.01,0.06430720090866089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,64,power_law_1.2,0.17085440158843995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,64,power_law_1.01,3.377388763427734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,64,power_law_1.01,0.05067520141601563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,64,power_law_1.2,0.17612799406051635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,64,power_law_1.2,0.19884159564971923
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,64,power_law_1.01,0.11666560173034668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,64,power_law_1.2,0.18600319623947142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,64,power_law_1.2,0.21171200275421143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,64,power_law_1.01,0.11942399740219116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,64,power_law_1.2,0.20034561157226563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,64,power_law_1.2,0.22548480033874513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,64,power_law_1.01,0.11832959651947021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,64,power_law_1.2,0.2147455930709839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,64,power_law_1.2,0.24827520847320556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,64,power_law_1.01,0.0931007981300354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,32,1,power_law_1.01,13.615321350097656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,64,power_law_1.2,0.29319679737091064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,64,power_law_1.01,0.08809599876403809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,64,power_law_1.01,4.416876983642578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,64,power_law_1.2,0.29991679191589354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,64,power_law_1.2,0.3504640102386475
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,64,power_law_1.01,0.0667136013507843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,64,power_law_1.2,0.38177919387817383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,64,power_law_1.2,0.4327040195465088
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,64,power_law_1.01,0.06503679752349853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,64,power_law_1.01,0.06547840237617493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,64,power_law_1.2,0.4540736198425293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,64,power_law_1.2,0.5086527824401855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,64,power_law_1.01,0.07093120217323304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,32,1,power_law_1.2,4.585919952392578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,64,power_law_1.2,0.6015935897827148
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,64,power_law_1.01,0.08315520286560059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,64,power_law_1.2,0.690828800201416
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,64,power_law_1.01,0.09815679788589478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,64,power_law_1.2,0.24053120613098145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,64,power_law_1.2,0.7559103965759277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,64,power_law_1.2,0.8576064109802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,64,power_law_1.01,0.11943039894104004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,64,power_law_1.2,0.2809344053268433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,64,power_law_1.2,1.0689023971557616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,64,power_law_1.01,0.14127999544143677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,64,power_law_1.2,0.3179712057113647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,64,power_law_1.2,1.2138879776000977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,64,power_law_1.01,0.19167360067367553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,64,power_law_1.2,0.407532787322998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,64,power_law_1.2,1.3162303924560548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,64,power_law_1.01,0.2440000057220459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,32,1,power_law_1.2,5.274208068847656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,64,power_law_1.2,0.48533120155334475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,64,power_law_1.2,1.640620803833008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,64,power_law_1.01,0.34741759300231934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,64,power_law_1.2,0.6511616230010986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,64,power_law_1.2,1.9819007873535157
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,64,power_law_1.01,0.46161279678344724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,64,power_law_1.2,0.8198847770690918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,64,power_law_1.01,0.6739136219024658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,64,power_law_1.2,2.4779775619506834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,64,power_law_1.2,2.6835840225219725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,64,power_law_1.2,1.1896448135375977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,64,power_law_1.01,0.8897791862487793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,64,balanced,0.04182399809360504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,32,1,power_law_1.2,5.927084732055664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,64,balanced,0.043680002291997276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,64,balanced,0.043978666265805565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,64,balanced,0.045738667249679565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,64,balanced,0.06558399895826976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,64,balanced,0.09909333785374959
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,64,power_law_1.2,0.03524479866027832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,64,balanced,0.10451733072598775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,64,balanced,0.10684266686439514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,64,balanced,0.11711999773979187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,64,balanced,0.11739200353622437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,32,1,power_law_1.01,19.010041809082033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,64,power_law_1.2,0.03603839874267578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,64,balanced,0.15843733151753744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,64,balanced,0.16449600458145142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,64,balanced,0.16914665699005127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,64,power_law_1.2,0.03591679930686951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,64,balanced,0.17691733439763388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,64,balanced,0.1832586725552877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,64,power_law_1.2,4.065222549438476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,64,balanced,0.19365866978963217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,64,power_law_1.2,0.03540480136871338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,64,balanced,0.20862932999928793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,64,balanced,0.2529386679331462
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,64,balanced,0.2818453311920166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,64,power_law_1.2,0.03548159897327423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,64,balanced,0.33588266372680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,64,balanced,0.3904533386230469
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,64,power_law_1.2,0.038150399923324585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,64,balanced,0.5009599924087524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,64,balanced,0.6121866703033447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,64,power_law_1.2,0.05805439949035644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,32,1,power_law_1.2,7.2569023132324215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,64,power_law_1.2,3.1889215469360352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,64,balanced,0.0397119993964831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,64,balanced,0.872549295425415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,64,power_law_1.2,0.0471231997013092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,64,balanced,0.03782399992148081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,64,balanced,0.03989866624275843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,64,balanced,0.037946666280428566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,64,balanced,0.04797333478927612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,64,balanced,1.0965920289357503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,64,power_law_1.2,0.11727360486984253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,64,power_law_1.2,5.115264129638672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,64,balanced,0.05483733117580414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,64,balanced,0.056202664971351624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,64,balanced,0.05619200070699056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,64,balanced,1.5706772804260254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,64,balanced,0.055861334005991616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,64,balanced,0.06840533514817555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,64,balanced,0.09433600306510925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,64,balanced,0.09484799702962239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,64,balanced,0.09612266222635905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,64,balanced,0.17281599839528403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,64,balanced,2.046559969584147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,64,balanced,0.17459734280904135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,64,power_law_1.2,1.4671808242797852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,64,balanced,0.18067200978597006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,64,balanced,0.18786666790644327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,64,balanced,0.19412267208099365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,64,balanced,0.2064639925956726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,64,balanced,0.21553067366282144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,64,balanced,0.07445333401362102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,64,balanced,0.2291626731554667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,64,balanced,0.08782399694124858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,64,balanced,0.253546675046285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,64,balanced,0.0888266662756602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,64,balanced,0.09364266196886699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,64,balanced,0.30058666070302326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,64,power_law_1.2,0.11832959651947021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,64,power_law_1.2,2.2102848052978517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,64,balanced,0.34704001744588214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,32,1,power_law_1.2,8.588563537597656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,64,balanced,0.439242680867513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,64,power_law_1.2,0.09715840220451355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,64,balanced,0.5324853261311849
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,64,power_law_1.2,0.09178239703178406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,64,balanced,0.7521653175354004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,64,power_law_1.2,0.08150399923324585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,64,balanced,0.9347840150197347
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,64,power_law_1.2,0.06803839802742004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,64,power_law_1.2,2.9282047271728517
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,64,power_law_1.2,0.06348800063133239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,64,balanced,1.3355520566304524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,64,power_law_1.01,0.07557119727134705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,64,power_law_1.2,0.0648256003856659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,64,power_law_1.01,0.0749504029750824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,64,power_law_1.2,0.07303680181503296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,64,balanced,1.733290672302246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,64,balanced,0.09510933359464009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,64,balanced,0.14827199776967367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,64,balanced,0.15839466452598572
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,64,power_law_1.2,0.08688639998435974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,64,balanced,0.16313067078590393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,64,balanced,0.17272533973058066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,64,balanced,0.1811786691347758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,64,power_law_1.2,0.09827200174331666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,64,balanced,0.1911626656850179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,64,balanced,2.5327626864115396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,64,balanced,0.20990399519602457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,64,power_law_1.2,0.12042880058288574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,64,balanced,0.24829334020614624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,64,balanced,0.27889599402745563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,64,power_law_1.2,0.1476096034049988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,64,balanced,0.3397119839986165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,32,1,power_law_1.01,24.396473693847657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,64,power_law_1.2,0.20493440628051757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,64,balanced,3.3308213551839194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,64,power_law_1.01,0.06316800117492676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,64,power_law_1.2,0.2589695930480957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,32,1,power_law_1.2,11.209510040283202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,64,power_law_1.01,0.07575039863586426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,64,balanced,0.027765333652496338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,64,power_law_1.01,0.06304640173912049
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,64,balanced,0.029557332396507263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,64,balanced,0.0295413335164388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,64,power_law_1.01,0.07463039755821228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,64,balanced,0.03032533327738444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,64,power_law_1.01,0.07823359966278076
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,64,balanced,0.03148799886306127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,64,balanced,0.04011200120051702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,64,balanced,0.04685866832733154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,64,power_law_1.01,0.07741439938545228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,64,power_law_1.01,0.07948160171508789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,64,power_law_1.01,0.07604479789733887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,64,power_law_1.01,0.08023679852485657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,64,power_law_1.01,0.08596479892730713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,64,power_law_1.01,0.07747840285301208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,64,balanced,0.40250666936238605
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,64,power_law_1.01,0.10006400346755981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,64,balanced,0.5252639849980673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,64,power_law_1.01,0.10375679731369018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,64,balanced,0.648581345876058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,64,power_law_1.01,0.11498240232467652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,64,balanced,0.9162346522013346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,64,power_law_1.2,0.39378559589385986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,64,power_law_1.01,0.1156991958618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,64,power_law_1.01,0.0627776026725769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,64,balanced,1.1601866881052654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,64,power_law_1.2,0.5039487838745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,64,balanced,0.06012799839178721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,64,power_law_1.01,0.06371200084686279
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,64,balanced,0.11718933780988057
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,64,balanced,0.11664000153541565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,64,power_law_1.01,0.05985919833183288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,64,balanced,0.10564800103505452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,64,balanced,1.6699466705322266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,64,power_law_1.2,0.7533184051513672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,64,power_law_1.01,0.06686720252037048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,64,power_law_1.01,0.08453119993209839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,64,power_law_1.01,0.08702719807624817
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,64,power_law_1.2,0.9767680168151855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,64,balanced,2.176431973775228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,64,power_law_1.01,0.08702080249786377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,64,power_law_1.01,0.09505919814109802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,32,1,power_law_1.01,0.7008255958557129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,64,power_law_1.01,0.09181439876556396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,64,power_law_1.01,0.0955456018447876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,32,1,power_law_1.2,13.902195739746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,64,power_law_1.01,0.08903679847717286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,32,1,power_law_1.01,0.7576064109802246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,64,power_law_1.01,0.16968319416046143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,64,power_law_1.01,0.15530879497528077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,64,power_law_1.01,0.14632960557937622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,64,power_law_1.01,0.1732800006866455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,64,power_law_1.01,0.163372802734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,32,1,power_law_1.01,0.8729920387268066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,64,power_law_1.01,0.15626879930496215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,64,power_law_1.01,0.17994240522384644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,64,power_law_1.01,0.16609280109405516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,64,power_law_1.01,0.16154240369796752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,64,power_law_1.01,0.17248640060424805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,64,power_law_1.01,0.18614399433135986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,64,power_law_1.01,0.16892800331115723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,64,power_law_1.01,0.18315520286560058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,64,power_law_1.01,0.18606719970703126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,64,power_law_1.01,0.17921919822692872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,64,power_law_1.01,0.19249919652938843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,64,power_law_1.01,0.20332159996032714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,64,power_law_1.01,0.18985600471496583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,64,power_law_1.2,0.07523199915885925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,64,power_law_1.01,0.22017281055450438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,64,power_law_1.01,0.21553919315338135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,64,power_law_1.01,0.21282560825347902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,64,power_law_1.2,0.07452800273895263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,64,power_law_1.01,0.26301438808441163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,64,power_law_1.01,0.22286720275878907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,64,power_law_1.01,0.25114240646362307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,64,power_law_1.2,0.07527679800987244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,64,power_law_1.01,0.3029247999191284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,64,power_law_1.01,0.26642560958862305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,64,power_law_1.01,0.2832256078720093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,64,power_law_1.2,0.07317119836807251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,64,power_law_1.01,0.3705087900161743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,64,power_law_1.01,0.33194880485534667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,64,power_law_1.01,0.35929598808288576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,64,power_law_1.2,0.07935360074043274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,32,1,power_law_1.01,0.9904959678649903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,64,power_law_1.01,0.4476736068725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,64,power_law_1.01,0.4247744083404541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,64,power_law_1.2,0.08593279719352723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,64,power_law_1.01,0.5758399963378906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,64,power_law_1.01,0.5684864044189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,32,1,power_law_1.01,1.2322367668151855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,64,power_law_1.01,0.7518976211547852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,64,power_law_1.01,0.7139647960662842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,32,1,power_law_1.01,1.4522687911987304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,64,power_law_1.01,0.03323520123958588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,64,power_law_1.01,1.0151679992675782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,64,power_law_1.01,0.9997183799743652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,64,balanced,0.09705600142478943
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,64,balanced,0.08527466654777527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,64,balanced,0.06838933130105336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,64,balanced,0.06842666864395142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,64,power_law_1.01,0.39122560024261477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,64,balanced,0.061568001906077065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,64,power_law_1.01,1.2858495712280273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,64,power_law_1.01,1.2648639678955078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,64,balanced,0.06679999828338623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,64,balanced,0.07524799803892772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,64,power_law_1.01,0.5123968124389648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,64,power_law_1.2,0.09953280091285706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,64,power_law_1.2,0.10366079807281495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,64,power_law_1.01,0.6279168128967285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,64,power_law_1.01,1.8625152587890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,64,power_law_1.2,0.11403520107269287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,64,power_law_1.01,0.9099455833435058
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,64,power_law_1.01,0.03232640027999878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,32,1,power_law_1.01,1.9514047622680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,64,power_law_1.01,1.171232032775879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,64,power_law_1.01,0.03402239978313446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,64,power_law_1.01,2.5373952865600584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,64,power_law_1.01,0.034867200255393985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,64,power_law_1.2,0.06234239935874939
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,64,power_law_1.01,0.03200640082359314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,64,power_law_1.01,1.6074815750122071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,64,power_law_1.2,0.06339200139045716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,32,1,power_law_1.01,2.419910430908203
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,64,power_law_1.01,0.03596799969673157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,64,power_law_1.2,0.06225919723510742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,64,power_law_1.01,1.8262144088745118
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,64,power_law_1.01,0.04092159867286682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,64,power_law_1.2,0.06384639739990235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,64,power_law_1.2,0.11664639711380005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,64,power_law_1.01,2.1357696533203123
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,64,power_law_1.01,0.055206400156021115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,64,power_law_1.2,0.05901439785957337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,64,power_law_1.01,0.10774400234222412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,64,power_law_1.2,0.0680191993713379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,32,1,power_law_1.01,3.395232009887695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,64,power_law_1.01,2.4619903564453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,64,power_law_1.01,0.10616320371627808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,64,power_law_1.2,0.08709759712219238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,64,balanced,0.0843946635723114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,64,power_law_1.2,0.07501440048217774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,64,power_law_1.01,0.09662719964981079
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,64,balanced,0.10153599580128987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,64,power_law_1.2,0.08654720187187195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,64,balanced,0.11962667107582092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,64,power_law_1.01,3.1364864349365233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,64,power_law_1.01,0.08062719702720642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,64,power_law_1.2,0.07791360020637512
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,64,balanced,0.15446933110555014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,64,power_law_1.2,0.0892799973487854
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,64,balanced,0.19158933560053507
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,64,power_law_1.01,0.07953919768333435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,64,power_law_1.2,0.07856640219688416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,64,power_law_1.2,0.09026560187339783
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,64,balanced,0.2731413245201111
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,64,power_law_1.01,0.0643775999546051
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,64,balanced,0.34623467922210693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,64,power_law_1.2,0.14583679437637329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,64,power_law_1.01,0.0638592004776001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,64,power_law_1.2,0.15697280168533326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,64,balanced,0.49050132433573407
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,64,power_law_1.01,0.059961599111557004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,64,balanced,0.6376373370488485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,64,power_law_1.01,0.060575997829437254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,64,power_law_1.01,4.1328575134277346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,64,power_law_1.01,0.07429119944572449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,64,balanced,0.029232000311215717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,64,balanced,0.029189333319664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,64,balanced,0.029253333806991577
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,64,power_law_1.01,0.08736640214920044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,64,balanced,0.029487999776999157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,64,balanced,0.029215998947620392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,64,balanced,0.03133333226044973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,64,balanced,0.03541333228349686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,64,power_law_1.01,0.10691839456558228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,64,balanced,0.035605333745479584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,64,balanced,0.03775999943415324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,64,balanced,0.03845866769552231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,64,balanced,0.04390400151411692
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,64,power_law_1.01,0.12691839933395385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,64,balanced,0.04543466866016388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,64,balanced,0.04785599807898203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,64,balanced,0.04791999856630961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,64,power_law_1.01,0.17017600536346436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,32,1,power_law_1.01,4.348620986938476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,64,power_law_1.2,0.15650559663772584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,64,balanced,0.027424000203609467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,64,balanced,0.027493332823117573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,64,balanced,0.02743999908367793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,64,power_law_1.2,0.16109440326690674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,64,power_law_1.01,0.21898880004882812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,64,power_law_1.2,0.16996480226516725
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,64,power_law_1.01,0.316812801361084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,64,power_law_1.2,0.18108160495758058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,64,power_law_1.2,0.19051519632339478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,64,power_law_1.2,0.22273919582366944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,64,power_law_1.2,0.25557119846343995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,64,power_law_1.2,0.07759360074996949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,64,power_law_1.2,0.29050240516662595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,32,1,power_law_1.2,19.187590026855467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,64,power_law_1.2,0.1629696011543274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,64,balanced,0.02741866558790207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,64,power_law_1.2,0.37079041004180907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,64,balanced,0.029616000751654308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,64,balanced,0.029535998900731403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,64,balanced,0.03350399931271871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,64,power_law_1.2,0.16802560091018676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,64,balanced,0.0334346666932106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,64,balanced,0.037818667789300285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,64,power_law_1.2,0.4386496067047119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,64,balanced,0.037578667203585304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,64,power_law_1.2,0.1751360058784485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,64,power_law_1.01,0.40804481506347656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,64,power_law_1.2,0.18513920307159423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,64,power_law_1.2,0.604256010055542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,64,power_law_1.2,0.19614720344543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,64,power_law_1.01,0.5886784076690674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,64,power_law_1.2,0.03288959860801697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,64,power_law_1.2,0.2273024082183838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,64,power_law_1.2,0.03328000009059906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,64,power_law_1.2,0.2678272008895874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,64,power_law_1.2,0.07891839742660522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,64,power_law_1.2,0.032390400767326355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,64,power_law_1.2,0.31383678913116453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,64,power_law_1.2,0.0869376003742218
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,64,power_law_1.2,0.031385600566864014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,64,power_law_1.2,0.38988800048828126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,64,power_law_1.2,0.09230080246925354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,64,balanced,0.0458133320013682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,64,balanced,0.04796266555786133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,64,power_law_1.2,0.03372800052165985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,64,balanced,0.04990399877230326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,64,power_law_1.2,0.47377920150756836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,64,power_law_1.2,0.09559040069580078
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,64,power_law_1.2,0.036575999855995175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,64,power_law_1.2,0.16967040300369263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,64,power_law_1.2,0.62740478515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,64,power_law_1.2,0.736729621887207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,64,power_law_1.2,0.044038400053977966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,64,power_law_1.2,0.1720639944076538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,64,power_law_1.2,0.767795181274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,64,power_law_1.2,0.05419520139694214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,64,power_law_1.2,0.17781120538711548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,64,power_law_1.01,0.7812543869018554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,64,power_law_1.2,1.0522432327270508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,64,power_law_1.2,0.18568960428237916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,64,power_law_1.2,1.1210047721862793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,64,power_law_1.2,1.332249641418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,64,balanced,0.053818667928377785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,64,power_law_1.2,1.4440064430236816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,64,power_law_1.2,1.9753984451293944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,64,balanced,0.03575466573238373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,64,power_law_1.2,2.0821823120117187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,64,balanced,0.035317334036032356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,64,power_law_1.2,0.1094655990600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,64,balanced,0.03571200122435888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,64,balanced,0.03751466671625773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,64,balanced,0.0356480007370313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,64,balanced,0.037674665451049805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,64,power_law_1.2,0.19151359796524048
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,64,power_law_1.2,0.09992960095405579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,64,power_law_1.2,0.199891197681427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,64,power_law_1.2,2.5808639526367188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,64,power_law_1.2,0.0917631983757019
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,64,power_law_1.2,0.22459518909454346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,64,power_law_1.2,0.08940799832344055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,64,power_law_1.2,0.2385792016983032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,64,balanced,0.0540533314148585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,64,power_law_1.2,0.0745855987071991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,64,balanced,0.05884266893068949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,64,balanced,0.06329600016276042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,64,power_law_1.2,0.2730432033538818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,64,balanced,0.07333333293596904
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,64,power_law_1.2,0.06154239773750305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,64,power_law_1.2,0.3384255886077881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,64,power_law_1.2,0.06339840292930603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,64,power_law_1.2,0.4001664161682129
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,64,power_law_1.2,0.05664640069007874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,64,power_law_1.2,0.0631168007850647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,64,balanced,0.023210667073726654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,64,power_law_1.2,0.5426688194274902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,64,balanced,0.0233599990606308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,64,balanced,0.02345066765944163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,64,power_law_1.2,0.0797439992427826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,64,balanced,0.023386667172114056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,64,balanced,0.025455998877684276
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,64,balanced,0.027306665976842243
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,64,power_law_1.2,0.0906112015247345
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,64,balanced,0.03379733363787333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,64,balanced,0.035786665976047516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,64,power_law_1.2,2.9828800201416015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,64,balanced,0.035616000493367515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,64,power_law_1.2,0.11013760566711425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,64,power_law_1.01,0.02748799920082092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,64,balanced,0.08550399541854858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,64,balanced,0.11107200384140015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,64,power_law_1.01,0.02728320062160492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,64,balanced,0.13155733545621237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,64,balanced,0.1844586730003357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,64,power_law_1.01,0.027724799513816834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,64,balanced,0.22985066970189413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,64,balanced,0.32075732946395874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,64,power_law_1.01,0.027718400955200194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,64,balanced,0.4140373468399048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,64,balanced,0.039450667798519135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,64,balanced,0.03973866750796636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,64,balanced,0.5997066497802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,64,balanced,0.05216533442338308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,64,balanced,0.05277866621812185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,64,balanced,0.05357866485913595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,64,balanced,0.783946673075358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,64,balanced,0.056746666630109154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,64,balanced,0.05552533268928528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,64,power_law_1.2,0.6677951812744141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,64,power_law_1.01,0.02794879972934723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,64,balanced,0.03664533297220866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,64,power_law_1.01,0.02773120105266571
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,64,balanced,0.03714133302370707
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,64,balanced,0.03945599993069967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,64,power_law_1.2,0.9742400169372558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,64,balanced,0.0394400010506312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,64,power_law_1.01,0.027564799785614012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,64,power_law_1.2,0.1398848056793213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,64,balanced,0.03606933355331421
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,64,balanced,0.03774400055408478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,64,power_law_1.01,0.02770560085773468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,64,balanced,0.03745066622893015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,64,balanced,0.03745600084463755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,64,power_law_1.2,0.18577280044555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,64,balanced,0.039877332746982574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,64,power_law_1.01,0.028646400570869444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,64,balanced,0.04354666670163473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,64,power_law_1.2,1.251583957672119
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,64,balanced,0.0481333335240682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,64,power_law_1.01,0.030067199468612672
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,64,balanced,0.052832002441088356
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,64,power_law_1.2,0.2334656000137329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,64,balanced,0.0633493314186732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,64,power_law_1.01,0.02794879972934723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,64,power_law_1.01,0.032915198802948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,64,balanced,0.07471466561158498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,64,power_law_1.01,0.03033599853515625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,64,balanced,0.10098666946093242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,64,power_law_1.01,0.033395200967788696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,64,power_law_1.2,0.34097919464111326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,64,balanced,0.13395733634630838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,64,power_law_1.01,0.033030399680137636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,64,power_law_1.01,0.037350401282310486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,64,balanced,0.19134400288263956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,64,power_law_1.2,1.7359807968139649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,64,balanced,0.06122133135795593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,64,power_law_1.01,0.03353599905967712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,64,balanced,0.06218666831652323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,64,balanced,0.2504746715227763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,64,power_law_1.2,0.45000319480895995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,64,balanced,0.06878933310508728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,32,1,power_law_1.2,24.576658630371092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,64,balanced,0.07632533212502797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,64,power_law_1.01,0.036774399876594546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,64,balanced,0.0950933297475179
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,64,balanced,0.11687999963760376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,64,power_law_1.01,0.037145599722862244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,64,balanced,0.15217600266138712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,64,balanced,0.18888533115386963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,64,power_law_1.01,0.04291839897632599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,64,balanced,0.26742400725682575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,64,power_law_1.01,0.04423680007457733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,64,balanced,0.3415520191192627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,64,power_law_1.2,2.352128028869629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,64,power_law_1.01,0.0455487996339798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,64,balanced,0.48813335100809735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,64,power_law_1.01,0.047635200619697574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,64,balanced,0.6342986822128296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,64,balanced,0.9304107030232748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,64,balanced,1.2217600345611572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,64,power_law_1.01,0.03768320083618164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,64,power_law_1.01,0.02242559939622879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,64,power_law_1.01,0.04446719884872437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,64,power_law_1.2,0.02650879919528961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,64,balanced,0.05811200042565664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,64,balanced,0.06230400005976359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,64,power_law_1.01,0.04785920083522797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,64,power_law_1.2,0.026438400149345398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,64,power_law_1.2,0.6402624130249024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,64,power_law_1.01,0.04892799854278564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,64,power_law_1.2,0.026688000559806822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,64,power_law_1.01,0.052070397138595584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,64,power_law_1.2,0.027423998713493346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,64,power_law_1.2,0.8330752372741699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,64,power_law_1.2,0.027596798539161683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,64,power_law_1.2,0.02956160008907318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,64,power_law_1.2,0.026598399877548216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,64,power_law_1.2,0.032793599367141726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,64,power_law_1.2,0.026643198728561402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,64,power_law_1.2,0.03280639946460724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,64,power_law_1.2,0.036057600378990175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,64,power_law_1.2,0.02710399925708771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,64,power_law_1.2,0.03648000061511993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,64,power_law_1.2,0.027577599883079527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,64,power_law_1.2,3.4303424835205076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,64,power_law_1.2,0.04258559942245484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,64,power_law_1.01,0.03468799889087677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,64,power_law_1.2,0.028384000062942505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,64,balanced,0.06840000053246816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,64,power_law_1.2,0.0434688001871109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,64,balanced,0.07667733232180278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,64,power_law_1.01,0.03485440015792847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,64,power_law_1.2,0.045516800880432126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,64,power_law_1.01,0.035155200958251955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,64,power_law_1.2,0.047839999198913574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,64,power_law_1.01,0.034764799475669864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,64,power_law_1.01,0.035462400317192076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,64,power_law_1.01,0.03759360015392303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,64,power_law_1.01,0.038047999143600464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,64,power_law_1.2,4.525535964965821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,64,power_law_1.01,0.038540801405906676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,64,power_law_1.01,0.04928640127182007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,64,power_law_1.2,0.03576320111751556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,64,power_law_1.01,0.02248319983482361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,64,power_law_1.01,0.05223039984703064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,64,power_law_1.2,0.03607040047645569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,64,power_law_1.2,0.029843199253082275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,64,power_law_1.01,0.052928000688552856
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,64,power_law_1.01,0.022035199403762817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,64,power_law_1.2,0.03589119911193848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,64,power_law_1.2,0.032441601157188416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,64,power_law_1.01,0.053350400924682614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,64,power_law_1.01,0.02252800017595291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,64,power_law_1.2,0.032927998900413515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,64,power_law_1.01,0.054790401458740236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,64,power_law_1.2,0.037036800384521486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,64,power_law_1.01,0.024428799748420715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,64,power_law_1.01,0.060787200927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,64,power_law_1.2,0.03739520013332367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,64,power_law_1.01,0.026662400364875792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,64,power_law_1.01,0.06465920209884643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,64,power_law_1.2,0.04402559995651245
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,64,power_law_1.01,0.03308799862861633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,64,power_law_1.01,0.0705344021320343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,64,power_law_1.2,0.04730879962444305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,64,power_law_1.01,0.03436799943447113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,64,power_law_1.2,0.04873600006103516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,64,power_law_1.01,0.03457280099391937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,64,power_law_1.2,0.051583999395370485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,64,power_law_1.01,0.03475199937820435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,64,balanced,0.09284800291061401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,64,balanced,0.1202826698621114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,64,power_law_1.01,0.03505919873714447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,64,balanced,0.14614933729171753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,64,balanced,0.20772266387939453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,64,power_law_1.2,0.03614720106124878
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,64,power_law_1.01,0.03619199991226196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,64,balanced,0.2609813412030538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,64,power_law_1.2,0.03671680092811584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,64,power_law_1.01,0.03600000143051148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,64,balanced,0.36869335174560547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,64,power_law_1.2,0.03835520148277283
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,64,power_law_1.01,0.03523840010166168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,64,power_law_1.2,0.0385919988155365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,64,power_law_1.01,0.035622400045394895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,64,power_law_1.2,0.03943040072917938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,64,power_law_1.01,0.08017280101776122
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,64,power_law_1.01,0.03604480028152466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,64,power_law_1.2,0.05057280063629151
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,64,power_law_1.01,0.09997439980506898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,64,power_law_1.01,0.0359935998916626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,64,power_law_1.2,0.05198079943656921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,64,power_law_1.01,0.12213120460510254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,64,power_law_1.01,0.03916800022125244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,64,power_law_1.2,0.05335680246353149
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,64,power_law_1.01,0.04273920059204102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,64,power_law_1.2,0.05620480179786682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,64,power_law_1.01,0.04753279983997345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,64,power_law_1.2,0.056403201818466184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,64,power_law_1.01,0.05448960065841675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,64,power_law_1.2,0.06170240044593811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,64,balanced,0.47270401318868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,64,power_law_1.01,0.06918399930000305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,64,power_law_1.2,0.06615039706230164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,64,balanced,0.6843679745992025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,64,power_law_1.01,0.08513919711112976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,64,power_law_1.2,0.07095680236816407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,64,power_law_1.2,0.0800320029258728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,64,balanced,0.8948799769083658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,64,power_law_1.2,0.1017024040222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,64,power_law_1.2,0.12871040105819703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,64,power_law_1.01,0.162336003780365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,64,power_law_1.2,0.16965759992599488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,64,power_law_1.01,0.2081216096878052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,64,power_law_1.2,0.21443839073181153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,64,power_law_1.01,0.056595200300216676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,64,power_law_1.01,0.29869439601898196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,64,power_law_1.01,0.06157439947128296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,64,power_law_1.2,0.3039360046386719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,64,power_law_1.01,0.06874880194664001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,64,power_law_1.01,0.05375360250473023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,64,power_law_1.01,0.07796480059623719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,64,power_law_1.01,0.057734400033950806
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,64,power_law_1.2,0.022419199347496033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,64,power_law_1.01,0.12345600128173828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,64,power_law_1.01,0.09503359794616699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,64,power_law_1.01,0.06427519917488098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,64,power_law_1.2,0.02245119959115982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,64,power_law_1.01,0.07495679855346679
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,64,power_law_1.2,0.022144000232219695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,64,power_law_1.01,0.08991360068321227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,64,power_law_1.2,0.02295680046081543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,64,power_law_1.01,0.11463040113449097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,64,power_law_1.2,0.024281600117683412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,64,power_law_1.01,0.13777920007705688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,64,power_law_1.2,0.026815998554229736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,64,power_law_1.2,0.053235197067260744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,64,power_law_1.01,0.19701759815216063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,64,power_law_1.2,0.03322240114212036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,64,power_law_1.2,0.05932160019874573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,64,power_law_1.01,0.2508543968200684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,64,power_law_1.01,0.3791232109069824
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,64,power_law_1.2,0.034355199337005614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,64,power_law_1.2,0.06447359919548035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,64,power_law_1.2,0.03427839875221252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,64,power_law_1.01,0.3542335987091064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,64,power_law_1.01,0.5399744033813476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,64,power_law_1.2,0.03485440015792847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,64,power_law_1.01,0.4569087982177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,64,power_law_1.01,0.1577664017677307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,64,power_law_1.2,0.035411199927330016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,64,power_law_1.01,0.7186431884765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,64,power_law_1.2,0.035308799147605895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,64,power_law_1.01,0.21747200489044188
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,64,power_law_1.2,0.036083200573921205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,64,power_law_1.01,0.2906752109527588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,64,power_law_1.01,1.045081615447998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,64,power_law_1.2,0.03532800078392029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,64,balanced,0.035386666655540466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,64,balanced,0.03473600000143051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,64,balanced,0.03555200000603994
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,64,power_law_1.2,0.03607679903507233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,64,balanced,0.035605333745479584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,64,balanced,0.04080000023047129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,64,balanced,0.04385066529115041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,64,balanced,0.051551997661590576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,64,balanced,0.05256533126036326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,64,balanced,0.06638933221499126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,64,balanced,0.0680159976085027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,64,power_law_1.2,0.07582079768180847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,64,balanced,0.07881600161393483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,64,balanced,0.08373333017031352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,64,balanced,0.08843732873598735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,64,balanced,0.09390933314959209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,64,power_law_1.01,0.1253376007080078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,64,power_law_1.01,0.15453439950942993
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,64,power_law_1.01,0.677350378036499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,64,power_law_1.01,0.21676158905029297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,64,power_law_1.01,0.2753151893615723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,64,power_law_1.01,0.8744959831237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,64,power_law_1.2,0.05740799903869629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,64,balanced,0.033488000432650246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,64,balanced,0.033471999069054924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,64,power_law_1.2,0.0615231990814209
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,64,power_law_1.01,0.39162240028381345
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,64,power_law_1.2,0.03578880131244659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,64,power_law_1.2,0.06926079988479614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,64,power_law_1.01,0.5015359878540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,64,power_law_1.2,0.07812479734420777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,64,power_law_1.2,0.4031424045562744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,64,power_law_1.2,0.09512320160865784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,64,power_law_1.01,0.7285888195037842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,64,power_law_1.2,0.12661759853363036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,64,power_law_1.2,0.15365760326385497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,64,power_law_1.01,0.9501376152038574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,64,balanced,0.04808533191680908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,64,balanced,0.048112000028292336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,64,power_law_1.2,0.22006399631500245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,64,balanced,0.04755199948946635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,64,balanced,0.047594666481018066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,64,balanced,0.0476693312327067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,64,balanced,0.05970133344332377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,64,power_law_1.2,0.28053760528564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,64,balanced,0.05852800110975901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,64,balanced,0.057589332262674965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,64,balanced,0.033530667424201965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,64,balanced,0.035642666121323906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,64,balanced,0.03561066587766012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,64,balanced,0.039664000272750854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,64,balanced,0.04417600234349569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,64,balanced,0.04584000011285146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,64,power_law_1.2,0.036364799737930296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,64,balanced,0.050213331977526345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,64,balanced,0.05230933427810669
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,64,power_law_1.2,0.040320000052452086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,64,power_law_1.2,0.09207040071487427
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,64,power_law_1.2,0.04293760061264038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,64,power_law_1.2,0.11623680591583252
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,64,power_law_1.2,0.05011839866638183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,64,power_law_1.2,0.14999680519104003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,64,power_law_1.2,0.060127997398376466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,64,power_law_1.2,0.20454399585723876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,64,power_law_1.2,0.2580159902572632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,64,balanced,0.0842026670773824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,64,balanced,0.0846666693687439
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,64,balanced,0.08769067128499348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,64,balanced,0.09572266538937886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,64,power_law_1.01,1.4016575813293457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,64,balanced,0.10363733768463135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,64,balanced,0.11460799972216289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,64,balanced,0.12310933073361714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,64,balanced,0.06842133402824402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,64,balanced,0.13064000010490417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,64,balanced,0.14564800262451172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,64,balanced,0.1739199956258138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,64,balanced,0.2053813338279724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,64,balanced,0.27938665946324664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,64,balanced,0.3466240167617798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,64,balanced,0.4857120116551717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,64,balanced,0.6263786554336548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,64,balanced,0.9068266550699869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,64,power_law_1.2,0.3940927982330322
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,64,power_law_1.2,0.07255679965019227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,64,balanced,1.1848266919453938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,64,power_law_1.2,0.512275218963623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,64,balanced,0.07450133562088013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,64,balanced,0.07838933169841766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,64,balanced,0.0848479966322581
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,64,balanced,0.025253333151340485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,64,balanced,0.025072000920772552
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,64,balanced,0.02517866591612498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,64,balanced,0.02534399926662445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,64,balanced,0.02937600016593933
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,64,balanced,0.03164266546567281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,64,balanced,0.03976000100374222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,64,power_law_1.2,0.09091839790344239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,64,balanced,0.04090133309364319
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,64,balanced,0.05799466868241628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,64,balanced,0.0617439995209376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,64,balanced,0.04078399886687597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,64,balanced,1.7440800666809082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,64,balanced,0.0415786678592364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,64,balanced,0.1035093367099762
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,64,balanced,0.04196266829967499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,64,balanced,0.10958400368690491
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,64,balanced,0.041690667470296226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,64,balanced,0.1209440032641093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,64,balanced,0.13796266913414001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,64,balanced,0.15734933813412985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,64,balanced,0.1962453325589498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,64,balanced,2.3000052769978843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,64,balanced,0.23003733158111572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,64,balanced,0.3181973298390706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,64,power_law_1.01,0.03727999925613403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,64,balanced,0.3973280191421509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,64,power_law_1.01,0.03726719915866852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,64,balanced,0.5634186665217081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,64,power_law_1.01,0.03712640106678009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,64,balanced,0.7299253145853678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,64,power_law_1.01,0.038124799728393555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,64,power_law_1.01,0.0394239991903305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,64,balanced,1.0629119873046875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,64,power_law_1.01,0.041868799924850465
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,64,balanced,0.04576000074545542
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,64,power_law_1.2,0.13073920011520385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,64,balanced,0.045925334095954895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,64,balanced,1.3944427172342937
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,64,balanced,0.04569066564242045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,64,balanced,0.049786667029062905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,64,power_law_1.01,0.032358399033546446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,64,power_law_1.01,0.03307519853115082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,64,power_law_1.01,0.03353599905967712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,64,power_law_1.01,0.03383679986000061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,64,power_law_1.01,0.03561600148677826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,64,power_law_1.01,0.036774399876594546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,64,power_law_1.01,0.043289598822593686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,64,power_law_1.01,0.0488319993019104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,64,power_law_1.01,0.04408319890499115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,64,power_law_1.01,0.04991360008716583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,64,power_law_1.01,0.04960640072822571
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,64,balanced,0.05382933219273885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,64,power_law_1.01,0.05073919892311096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,64,balanced,0.06302399933338165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,64,balanced,0.07259733478228252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,64,power_law_1.01,0.06705920100212097
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,64,balanced,0.10513066252072652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,64,balanced,0.13172266880671182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,64,power_law_1.01,0.07285760045051574
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,64,balanced,0.1872319976488749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,64,power_law_1.01,0.07772160172462464
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,64,balanced,0.24378132820129395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,64,power_law_1.2,0.5708608150482177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,64,power_law_1.01,0.08511999845504761
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,64,balanced,0.34507731596628827
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,64,balanced,0.4470133384068807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,64,power_law_1.01,0.04506880044937134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,64,power_law_1.01,0.04522239863872528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,64,power_law_1.01,0.06236799955368042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,64,power_law_1.01,0.045203199982643126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,64,power_law_1.01,0.06334720253944397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,64,power_law_1.01,0.04544639885425568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,64,power_law_1.01,0.07505919933319091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,64,power_law_1.01,0.046214398741722104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,64,power_law_1.01,0.08144000172615051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,64,power_law_1.01,0.05690240263938904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,64,power_law_1.01,0.0887935996055603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,64,power_law_1.01,0.056569600105285646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,64,power_law_1.01,0.0970687985420227
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,64,power_law_1.2,0.17046400308609008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,64,power_law_1.01,0.0581055998802185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,64,power_law_1.01,0.10536960363388062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,64,power_law_1.01,0.08245760202407837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,64,power_law_1.01,0.11055359840393067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,64,power_law_1.01,0.08440960049629212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,64,power_law_1.01,0.12268799543380737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,64,power_law_1.01,0.09056000113487243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,64,balanced,0.09294399619102478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,64,power_law_1.01,0.13927040100097657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,64,balanced,0.09934932986895244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,64,power_law_1.01,0.09436799883842469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,64,power_law_1.01,0.1646399974822998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,64,power_law_1.01,0.10044159889221191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,64,power_law_1.01,0.2145087957382202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,64,power_law_1.01,0.11308159828186035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,64,power_law_1.2,0.7664063930511474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,64,power_law_1.2,0.3825727939605713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,64,power_law_1.01,0.2768768072128296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,64,power_law_1.01,0.1225600004196167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,64,power_law_1.01,0.13245439529418945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,64,power_law_1.01,0.15123840570449829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,64,power_law_1.01,0.18726400136947632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,64,power_law_1.01,0.23057920932769777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,64,balanced,0.11532800396283467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,64,power_law_1.01,0.32610559463500977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,64,balanced,0.13200533390045166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,64,balanced,0.15449066956837973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,64,balanced,0.20454933245976767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,64,power_law_1.01,0.4202752113342285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,64,balanced,0.2487679918607076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,64,balanced,0.3429066737492879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,64,power_law_1.01,0.5939455986022949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,64,balanced,0.43297068277994794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,64,power_law_1.01,0.3720576047897339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,64,balanced,0.6136426528294882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,64,power_law_1.01,0.7653823852539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,64,power_law_1.01,0.4880703926086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,64,balanced,0.7960533301035563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,64,power_law_1.01,0.6745664119720459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,64,balanced,1.1578400135040283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,64,power_law_1.01,1.1112832069396972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,64,power_law_1.01,0.8889535903930664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,64,balanced,1.5213707288106282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,64,power_law_1.01,1.43438720703125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,64,power_law_1.01,1.3086655616760254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,64,power_law_1.01,1.6978944778442382
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,64,power_law_1.2,0.03694719970226288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,64,power_law_1.2,0.03731839954853058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,64,power_law_1.2,0.03699199855327606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,64,power_law_1.2,0.03809280097484589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,64,power_law_1.2,0.03898879885673523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,64,power_law_1.2,0.04135040044784546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,64,power_law_1.2,0.485152006149292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,64,power_law_1.01,2.2404415130615236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,64,power_law_1.01,0.09470080137252808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,64,power_law_1.01,0.10136959552764893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,64,power_law_1.01,0.11799039840698242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,64,power_law_1.01,0.02563839852809906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,64,power_law_1.01,0.1349120020866394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,64,power_law_1.01,0.025753599405288697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,64,power_law_1.01,0.16087679862976073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,64,power_law_1.01,0.026150399446487428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,64,power_law_1.01,0.21697280406951905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,64,power_law_1.01,0.026547199487686156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,64,power_law_1.01,0.2709439992904663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,64,power_law_1.2,0.04915199875831604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,64,power_law_1.01,0.029075199365615846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,64,power_law_1.2,0.04996480047702789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,64,power_law_1.01,0.3700671911239624
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,64,power_law_1.01,0.03144319951534271
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,64,power_law_1.2,0.062009602785110474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,64,power_law_1.01,0.039155200123786926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,64,power_law_1.2,0.06963199973106385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,64,power_law_1.01,0.4803008079528809
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,64,power_law_1.01,0.04069119989871979
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,64,power_law_1.2,0.07475839853286743
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,64,power_law_1.01,0.05242879986763001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,64,power_law_1.2,0.08447999954223633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,64,power_law_1.01,0.6741888046264648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,64,power_law_1.01,0.051577597856521606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,64,power_law_1.2,0.08974720239639282
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,64,power_law_1.01,0.040300801396369934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,64,power_law_1.2,0.09743360280990601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,64,power_law_1.01,2.830022430419922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,64,power_law_1.01,0.8854911804199219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,64,power_law_1.2,0.10657279491424561
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,64,power_law_1.01,0.040166398882865904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,64,power_law_1.2,0.03270399868488312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,64,power_law_1.2,0.1123136043548584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,64,power_law_1.01,0.04087679982185364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,64,power_law_1.2,0.03301759958267212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,64,power_law_1.2,0.12522239685058595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,64,power_law_1.01,0.04041599929332733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,64,power_law_1.01,1.2740927696228028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,64,power_law_1.2,0.03285120129585266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,64,power_law_1.2,0.1457856059074402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,64,power_law_1.01,0.042342400550842284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,64,power_law_1.2,0.033657601475715636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,64,power_law_1.2,0.17765120267868043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,64,power_law_1.01,0.04407680034637451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,64,power_law_1.2,0.03489919900894165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,64,power_law_1.2,0.03699840009212494
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,64,power_law_1.01,0.045535999536514285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,64,power_law_1.01,1.7004032135009766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,64,power_law_1.2,0.04298880100250244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,64,power_law_1.01,0.051097601652145386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,64,power_law_1.2,0.04445439875125885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,64,power_law_1.2,0.04381439983844757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,64,power_law_1.01,0.058208000659942624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,64,power_law_1.2,0.04547199904918671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,64,power_law_1.01,0.07390720248222352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,64,power_law_1.2,0.04487679898738861
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,64,power_law_1.01,0.08886399865150452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,64,power_law_1.2,0.045952001214027406
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,64,power_law_1.01,0.12663040161132813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,64,power_law_1.2,0.046758401393890384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,64,power_law_1.01,0.15329920053482055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,64,power_law_1.2,0.05614079833030701
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,64,power_law_1.01,0.22047998905181884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,64,power_law_1.2,0.054048001766204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,64,power_law_1.2,0.23468799591064454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,64,power_law_1.2,0.05934079885482788
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,64,power_law_1.01,0.30280959606170654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,64,power_law_1.2,0.2950400114059448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,64,power_law_1.2,0.08109440207481385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,64,power_law_1.01,0.4136320114135742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,64,power_law_1.2,0.0827072024345398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,64,power_law_1.2,0.04999040067195892
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,64,power_law_1.2,0.08881919980049133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,64,power_law_1.2,0.05013759732246399
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,64,power_law_1.01,0.5489920139312744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,64,power_law_1.2,0.09560319781303406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,64,power_law_1.2,0.06638720035552978
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,64,power_law_1.2,0.24047999382019042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,64,power_law_1.2,0.10272639989852905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,64,power_law_1.2,0.07208319902420043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,64,power_law_1.2,0.0774399995803833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,64,power_law_1.2,0.1155135989189148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,64,power_law_1.2,0.08474239706993103
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,64,power_law_1.2,0.12570879459381104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,64,power_law_1.2,0.13479039669036866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,64,power_law_1.2,0.15217280387878418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,64,power_law_1.2,0.2071615934371948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,64,power_law_1.2,0.24990720748901368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,64,power_law_1.2,0.3562688112258911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,64,power_law_1.2,0.4585536003112793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,64,power_law_1.2,0.6391039848327636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,64,power_law_1.2,0.8259903907775878
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,64,power_law_1.2,0.025779199600219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,64,power_law_1.2,0.40253438949584963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,64,power_law_1.2,0.025696000456809996
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,64,power_law_1.2,0.025843200087547303
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,64,power_law_1.2,0.026815998554229736
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,64,power_law_1.2,0.029120001196861266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,64,power_law_1.2,0.03164800107479095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,64,power_law_1.2,0.03939839899539947
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,64,power_law_1.2,0.04038400053977966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,64,power_law_1.2,0.051923197507858274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,64,power_law_1.2,0.05145599842071533
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,64,power_law_1.2,0.03992320001125336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,64,power_law_1.2,0.040031999349594116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,64,power_law_1.2,0.04049279987812042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,64,power_law_1.2,0.04127359986305237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,64,power_law_1.2,0.043296000361442565
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,64,power_law_1.2,0.04410240054130554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,64,power_law_1.2,0.046316799521446225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,64,power_law_1.2,0.051974397897720334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,64,power_law_1.2,0.061337602138519284
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,64,power_law_1.2,0.0759552001953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,64,power_law_1.2,0.10269440412521362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,64,power_law_1.2,1.2151488304138183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,64,power_law_1.2,0.5116159915924072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,64,power_law_1.2,0.7039040088653564
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,64,power_law_1.2,0.13244160413742065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,64,power_law_1.2,0.09230719804763794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,64,power_law_1.2,0.10175360441207885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,64,power_law_1.2,0.11974400281906128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,64,power_law_1.2,0.7534719944000244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,64,power_law_1.2,0.1411967992782593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,64,power_law_1.2,0.16625920534133912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,64,power_law_1.2,0.22645120620727538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,64,power_law_1.2,0.2848959922790527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,64,power_law_1.2,0.37985279560089114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,64,power_law_1.2,1.1175168037414551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,64,power_law_1.2,0.5099775791168213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,64,power_law_1.2,0.17643519639968872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,64,power_law_1.2,0.7524479866027832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,64,power_law_1.2,1.6082815170288085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,64,power_law_1.2,0.3203392028808594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,64,balanced,0.05138133466243744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,64,balanced,0.05018133421738943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,64,power_law_1.2,0.7048831939697265
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,64,power_law_1.2,0.24461441040039061
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,64,power_law_1.2,0.9054719924926757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,64,power_law_1.2,0.30310399532318116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,64,balanced,0.0517546683549881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,64,balanced,0.04995200037956238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,64,balanced,0.0716480016708374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,64,balanced,0.11200533310572307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,64,balanced,0.1529973347981771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,64,balanced,0.155349334081014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,64,balanced,0.14493333299954733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,64,balanced,0.14735999703407288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,64,balanced,0.23726399739583334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,64,balanced,0.25226134061813354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,64,balanced,0.2581226627031962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,64,balanced,0.2674773335456848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,64,balanced,0.279146671295166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,64,balanced,0.28902933994929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,64,balanced,0.30662933985392254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,64,balanced,0.3738826513290405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,64,balanced,0.4033600091934204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,64,balanced,0.46374932924906415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,64,balanced,0.5245279868443807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,64,balanced,0.6480106512705485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,64,balanced,0.76910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,64,balanced,1.016858657201131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,64,balanced,1.3219467004140217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,64,balanced,1.8141973813374836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,64,balanced,2.3676746686299643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,64,balanced,0.039594667653242745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,64,balanced,0.041690667470296226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,64,balanced,0.04101866732041041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,64,power_law_1.2,0.9804351806640625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,64,balanced,0.041536000867684685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,64,balanced,0.054058666030565895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,64,balanced,0.0804746647675832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,64,balanced,0.12160000205039978
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,64,balanced,0.12376532951990764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,64,balanced,0.11297600467999776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,64,balanced,0.11691199739774068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,64,balanced,0.22666666905085245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,64,balanced,0.24067199230194092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,64,balanced,0.2482879956563314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,64,balanced,0.25674666961034137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,64,balanced,0.2690986593564351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,64,balanced,0.279258668422699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,64,balanced,0.3018239935239156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,64,balanced,0.35146665573120117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,64,balanced,0.3782186508178711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,64,balanced,0.44733333587646484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,64,balanced,0.5154293378194174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,64,balanced,0.6521600087483724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,64,balanced,0.7877600193023682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,64,power_law_1.2,1.4704511642456055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,64,balanced,1.056506633758545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,64,balanced,0.05836800237496694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,64,balanced,0.06028266747792562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,64,balanced,0.05995733539263407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,64,balanced,0.060032000144322716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,64,balanced,0.0743999977906545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,64,balanced,0.1015786627928416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,64,balanced,1.3615946769714355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,64,balanced,0.11450133721033733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,64,balanced,0.11926933129628499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,64,balanced,0.26521599292755127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,64,balanced,0.2699306607246399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,64,balanced,0.2760213414827983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,64,balanced,1.9030240376790364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,64,balanced,0.2834239999453227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,64,balanced,0.29029866059621173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,64,balanced,0.30394667387008667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,64,balanced,0.3192906578381856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,64,balanced,0.3298719922701518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,64,balanced,2.4732960065205893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,64,balanced,0.35782400767008465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,64,balanced,0.40562132994333905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,64,balanced,0.45716265837351483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,64,power_law_1.2,0.9866751670837403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,64,balanced,0.5564906597137451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,64,power_law_1.01,0.10711040496826171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,64,balanced,0.6560106674830118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,64,power_law_1.01,0.10744320154190064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,64,balanced,0.855066696802775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,64,power_law_1.01,0.10753920078277587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,64,balanced,1.1145599683125813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,64,power_law_1.01,0.10841599702835084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,64,balanced,1.5137440363566081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,64,power_law_1.2,0.9285696029663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,64,power_law_1.01,0.11009279489517212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,64,power_law_1.01,0.07542399764060974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,64,power_law_1.01,0.11500799655914307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,64,balanced,1.967850685119629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,64,power_law_1.01,0.07893760204315185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,64,power_law_1.01,0.1359488010406494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,64,power_law_1.01,0.07818880081176757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,64,power_law_1.01,0.14060159921646118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,64,balanced,2.8161706924438477
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,64,power_law_1.01,0.07982079982757569
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,64,power_law_1.01,0.1373247981071472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,64,power_law_1.01,0.07824640274047852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,64,power_law_1.01,0.14175360202789306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,64,balanced,0.029951999584833782
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,64,balanced,0.0308693324526151
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,64,balanced,0.031130666534105938
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,64,balanced,0.03153600047032038
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,64,balanced,0.03350933392842611
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,64,balanced,0.043477331598599754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,64,power_law_1.01,0.08353919982910156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,64,balanced,3.723679860432943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,64,power_law_1.2,1.3504704475402831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,64,power_law_1.01,0.09279360175132752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,64,power_law_1.01,0.09191039800643921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,64,power_law_1.01,0.23601920604705812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,64,power_law_1.01,0.09275519847869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,64,power_law_1.01,0.24539520740509033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,64,power_law_1.01,0.09353600144386291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,64,power_law_1.01,0.25182080268859863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,64,power_law_1.01,0.09144960045814514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,64,power_law_1.01,0.2601599931716919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,64,power_law_1.01,0.1118783950805664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,64,power_law_1.01,0.09955840110778809
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,64,power_law_1.01,0.27111680507659913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,64,power_law_1.01,0.11800960302352906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,64,power_law_1.01,0.11111680269241334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,64,power_law_1.01,0.2813823938369751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,64,power_law_1.01,0.10607359409332276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,64,power_law_1.01,0.3028543949127197
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,64,power_law_1.01,0.11223039627075196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,64,power_law_1.01,0.11138559579849243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,64,power_law_1.01,0.3678528070449829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,64,power_law_1.01,0.2522752046585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,64,power_law_1.01,0.22191359996795654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,64,balanced,0.05807466804981232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,64,power_law_1.01,0.39607040882110595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,64,balanced,0.06288533409436543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,64,power_law_1.01,0.2639552116394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,64,power_law_1.01,0.23395199775695802
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,64,balanced,0.08412266770998637
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,64,balanced,0.08646933237711589
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,64,balanced,0.20825600624084473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,64,power_law_1.01,0.47725439071655273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,64,power_law_1.01,0.2416383981704712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,64,power_law_1.01,0.26063361167907717
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,64,balanced,0.2097919980684916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,64,balanced,0.18131200472513834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,64,power_law_1.01,0.2503743886947632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,64,balanced,0.14550399780273438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,64,power_law_1.01,0.5620160102844238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,64,power_law_1.01,0.26997759342193606
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,64,balanced,0.08117866516113281
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,64,balanced,0.08285866677761078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,64,power_law_1.01,0.265068793296814
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,64,balanced,0.08694400389989217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,64,power_law_1.01,0.279584002494812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,64,power_law_1.01,0.7155712127685547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,64,balanced,0.09489066402117412
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,64,balanced,0.10365333159764607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,64,power_law_1.01,0.27302401065826415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,64,balanced,0.12150399883588155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,64,power_law_1.01,0.29996800422668457
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,64,balanced,0.140255997578303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,64,power_law_1.01,0.8747200012207031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,64,power_law_1.01,0.2978431940078735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,64,balanced,0.18143999576568604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,64,power_law_1.01,0.30450561046600344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,64,balanced,0.2243573268254598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,64,power_law_1.01,0.34669439792633056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,64,balanced,0.3095146616299947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,64,power_law_1.01,1.1760255813598632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,64,power_law_1.01,0.31217279434204104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,64,balanced,0.40302932262420654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,64,power_law_1.01,0.38057599067687986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,64,power_law_1.01,0.35489280223846437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,64,power_law_1.01,0.45457921028137205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,64,power_law_1.01,1.509113597869873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,64,power_law_1.01,0.41953282356262206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,64,power_law_1.01,0.5397568225860596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,64,power_law_1.01,0.49274239540100095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,64,power_law_1.01,0.6877503871917725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,64,power_law_1.01,2.120185661315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,64,power_law_1.01,0.6121088027954101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,64,power_law_1.01,0.8446016311645508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,64,power_law_1.01,0.7512383937835694
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,64,balanced,0.5574453274408976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,64,power_law_1.01,1.1450048446655274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,64,balanced,0.7424053351084391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,64,power_law_1.01,1.0162431716918945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,64,power_law_1.01,2.7091648101806642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,64,power_law_1.2,0.10668799877166749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,64,power_law_1.01,1.4437376022338868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,64,power_law_1.01,1.2969280242919923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,64,power_law_1.01,2.0647104263305662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,64,power_law_1.01,1.8262016296386718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,64,power_law_1.01,2.6847103118896483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,64,power_law_1.01,2.315545654296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,64,power_law_1.2,0.10714880228042603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,64,power_law_1.2,0.0757311999797821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,64,power_law_1.2,0.4532479763031006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,64,power_law_1.2,0.10825599431991577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,64,power_law_1.2,0.07906559705734253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,64,power_law_1.2,0.10789120197296143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,64,power_law_1.2,0.07826560139656066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,64,power_law_1.2,0.07970560193061829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,64,power_law_1.2,0.11370879411697388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,64,power_law_1.01,3.344723129272461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,64,power_law_1.2,0.08117120265960694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,64,power_law_1.2,0.11322239637374878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,64,power_law_1.2,0.08515840172767639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,64,power_law_1.2,0.13432960510253905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,64,power_law_1.2,0.1122048020362854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,64,power_law_1.2,0.14039679765701293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,64,power_law_1.01,0.04936319887638092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,64,power_law_1.2,0.11875200271606445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,64,power_law_1.2,0.13611520528793336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,64,power_law_1.01,4.42229118347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,64,power_law_1.01,0.04950399994850159
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,64,power_law_1.2,0.10535680055618286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,64,power_law_1.2,0.1388479948043823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,64,power_law_1.2,0.09141119718551635
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,64,power_law_1.01,0.04945279955863953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,64,power_law_1.2,0.10968960523605346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,64,power_law_1.2,0.2268928050994873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,64,power_law_1.2,0.09201279878616334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,64,power_law_1.01,0.04938240051269531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,64,power_law_1.2,0.21910400390625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,64,power_law_1.2,0.2427903890609741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,64,power_law_1.2,0.09232640266418457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,64,power_law_1.2,0.23133440017700196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,64,power_law_1.01,0.05073919892311096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,64,power_law_1.2,0.09344639778137206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,64,power_law_1.2,0.2475071907043457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,64,power_law_1.2,0.24245760440826417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,64,power_law_1.01,0.051481598615646364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,64,power_law_1.2,0.09445120096206665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,64,power_law_1.2,0.2518784046173096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,64,power_law_1.01,0.05130879878997803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,64,power_law_1.2,0.10210560560226441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,64,power_law_1.2,0.26165759563446045
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,64,power_law_1.01,0.058169597387313844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,64,power_law_1.2,0.27530879974365235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,64,power_law_1.2,0.11162240505218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,64,power_law_1.01,0.08082559704780579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,64,power_law_1.2,0.3070784091949463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,64,power_law_1.2,0.1106943964958191
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,64,power_law_1.01,0.07416319847106934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,64,power_law_1.2,0.3502079963684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,64,power_law_1.2,0.25905280113220214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,64,power_law_1.01,0.1850816011428833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,64,power_law_1.2,0.389516806602478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,64,power_law_1.2,0.26258559226989747
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,64,power_law_1.01,0.16119680404663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,64,power_law_1.2,0.47258877754211426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,64,power_law_1.2,0.26739840507507323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,64,power_law_1.01,0.15463680028915405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,64,power_law_1.2,0.5492032051086426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,64,power_law_1.2,0.2819008111953735
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,64,power_law_1.01,0.11470719575881957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,64,power_law_1.2,0.7151103973388672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,64,power_law_1.2,0.326092791557312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,64,power_law_1.01,0.07779200077056884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,64,power_law_1.2,0.36851840019226073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,64,power_law_1.01,0.08193280100822449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,64,power_law_1.2,0.8730175971984864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,64,power_law_1.2,0.419052791595459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,64,power_law_1.2,0.25540480613708494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,64,power_law_1.2,1.2133248329162598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,64,power_law_1.2,0.511187219619751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,64,power_law_1.2,0.26730880737304685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,64,power_law_1.2,0.5807168006896972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,64,power_law_1.2,1.503929615020752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,64,power_law_1.2,0.2763263940811157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,64,power_law_1.2,0.7284031867980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,64,power_law_1.2,0.281331205368042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,64,power_law_1.2,0.9058879852294922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,64,power_law_1.2,0.29269120693206785
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,64,power_law_1.01,0.08283519744873047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,64,power_law_1.2,2.201580810546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,64,power_law_1.2,0.3088320016860962
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,64,power_law_1.01,0.09160959720611572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,64,power_law_1.2,1.2654848098754883
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,64,power_law_1.01,0.09905279874801635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,64,power_law_1.2,0.323801589012146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,64,power_law_1.01,0.12245759963989258
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,64,power_law_1.2,0.35649919509887695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,64,power_law_1.2,1.634707260131836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,64,power_law_1.2,2.8313472747802733
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,64,power_law_1.01,0.144595205783844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,64,power_law_1.2,0.4350592136383057
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,64,power_law_1.01,0.19159679412841796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,64,power_law_1.2,0.48616318702697753
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,64,power_law_1.01,0.24673919677734374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,64,power_law_1.2,2.4416128158569337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,64,power_law_1.2,0.6386879920959473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,64,power_law_1.01,0.34714879989624026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,64,power_law_1.2,0.8070015907287598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,64,power_law_1.01,0.44827518463134763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,64,power_law_1.01,0.6316544055938721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,64,power_law_1.2,1.0785087585449218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,64,power_law_1.2,3.032454490661621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,balanced,0.03161599983771642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,balanced,0.033786666889985405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,balanced,0.03356799980004629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,balanced,0.033573334415753685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,balanced,0.035546667873859406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,balanced,0.056464001536369324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,balanced,0.05811200042565664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,64,power_law_1.2,1.344332790374756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,64,power_law_1.2,1.9258560180664062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,64,power_law_1.2,1.5319040298461915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,64,power_law_1.01,0.8830464363098145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,balanced,0.031199999153614044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,balanced,0.031221332649389904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,64,power_law_1.2,2.5467199325561523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,balanced,0.031248000760873158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,balanced,0.03315199911594391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,balanced,0.03342399994532267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,64,power_law_1.2,2.3870847702026365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,64,power_law_1.2,0.049516800045967105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,64,power_law_1.2,0.0490880012512207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,64,power_law_1.2,0.04919680058956146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,64,power_law_1.2,3.826598358154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,64,power_law_1.2,0.04872959852218628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,balanced,0.05844266712665558
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,64,power_law_1.2,0.04832000136375427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,balanced,0.06834133466084798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,balanced,0.07067733506361644
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,balanced,0.09285333752632141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,balanced,0.10114666819572449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,balanced,0.04133866727352142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,balanced,0.04786666731039683
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,64,power_law_1.2,0.048307201266288756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,balanced,0.04817600051561991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,balanced,0.04984533290068308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,balanced,0.049973333875338234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,balanced,0.09105066458384196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,balanced,0.09900800387064616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,64,power_law_1.2,0.048556798696517946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,balanced,0.10110400120417277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,balanced,0.10895466804504395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,balanced,0.11585066715876262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,balanced,0.12014933427174886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,balanced,0.13572800159454346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,balanced,0.15236799915631613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,balanced,0.17086400588353476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,64,power_law_1.2,5.078003311157227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,balanced,0.21182399988174438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,balanced,0.043663998444875084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,balanced,0.25442665815353394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,balanced,0.04193066557248434
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,balanced,0.044026667873064675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,balanced,0.3340640068054199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,balanced,0.043578664461771645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,balanced,0.045754666129748024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,balanced,0.05761066575845083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,balanced,0.40806933244069415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,balanced,0.057205334305763245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,balanced,0.05709333221117655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,balanced,0.5701013406117758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,balanced,0.10060800115267436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,64,power_law_1.2,1.7591552734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,balanced,0.10317333539326985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,balanced,0.10756799578666687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,balanced,0.7249813079833984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,balanced,0.11335466305414836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,balanced,0.11953066786130269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,balanced,0.12983466188112894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,balanced,0.15026666720708212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,balanced,1.0396373271942139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,balanced,0.16717867056528726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,balanced,0.2041226625442505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,balanced,0.23824000358581543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,64,power_law_1.2,0.05612159967422485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,balanced,0.30664533376693726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,balanced,1.3521226247151692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,balanced,0.3738186756769816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,64,power_law_1.2,0.0758463978767395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,power_law_1.01,0.043961599469184875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,balanced,0.5222400029500326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,balanced,0.6581279834111532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,power_law_1.01,0.044659200310707095
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,64,power_law_1.2,0.07400959730148315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,balanced,0.9443199634552002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,balanced,0.1029866635799408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,balanced,0.10732799768447876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,balanced,0.1114026705423991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,balanced,1.2232960065205891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,balanced,0.11353066563606262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,balanced,0.12383466958999634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,balanced,0.1306880017121633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,power_law_1.01,0.034585601091384886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,balanced,0.13937600453694662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,balanced,0.15525333086649576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,balanced,0.18288532892862955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,power_law_1.01,0.03467519879341126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,balanced,0.21312000354131064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,balanced,0.023503998915354412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,balanced,0.27350399891535443
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,balanced,0.02518400053183238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,balanced,0.025477332373460133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,power_law_1.01,0.0353983998298645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,balanced,0.3341226577758789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,balanced,0.46537065505981445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,power_law_1.01,0.03557760119438171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,balanced,0.5804106791814169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,power_law_1.01,0.037222400307655334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,power_law_1.01,0.044486400485038755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,balanced,0.8281599680582682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,power_law_1.01,0.040249601006507874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,64,power_law_1.2,0.17420159578323363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,power_law_1.01,0.04522239863872528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,balanced,1.070858637491862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,power_law_1.01,0.04754559993743897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,power_law_1.01,0.04180479943752289
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,64,power_law_1.2,0.140774405002594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,power_law_1.01,0.04791040122509003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,power_law_1.01,0.052876800298690796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,balanced,1.55949862798055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,64,power_law_1.2,0.1381824016571045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,power_law_1.01,0.04715520143508911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,power_law_1.01,0.054995197057724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,balanced,0.02550400048494339
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,power_law_1.01,0.04766719937324524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,power_law_1.01,0.05609599947929382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,balanced,0.027552001178264618
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,balanced,0.03129599988460541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,balanced,2.047013282775879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,balanced,0.033717334270477295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,power_law_1.01,0.09107199907302857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,power_law_1.01,0.06421759724617004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,balanced,0.035349334279696144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,balanced,0.04292800029118856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,power_law_1.01,0.04570879936218262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,balanced,0.04257066547870636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,power_law_1.01,0.06706560254096985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,balanced,0.06679999828338623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,balanced,0.05614933371543884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,power_law_1.01,0.04529919922351837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,power_law_1.01,0.09175040125846863
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,balanced,0.050010666251182556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,balanced,0.04472533365090688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,balanced,0.04569066564242045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,power_law_1.01,0.09796479940414429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,power_law_1.01,0.04535039961338043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,balanced,0.04744533201058706
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,balanced,0.047610665361086525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,power_law_1.01,0.10030720233917237
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,balanced,0.04778666794300079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,power_law_1.01,0.04593920111656189
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,balanced,0.0499893327554067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,power_law_1.01,0.10600320100784302
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,balanced,0.060085331400235496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,balanced,0.06854933500289917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,power_law_1.01,0.11207679510116578
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,balanced,0.08892800410588582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,balanced,0.11388799548149109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,power_law_1.01,0.11745920181274414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,64,power_law_1.2,0.12298239469528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,balanced,0.16064533591270447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,power_law_1.01,0.12899199724197388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,balanced,0.2020053267478943
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,64,power_law_1.2,0.07207679748535156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,balanced,0.28735466798146564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,power_law_1.01,0.15240319967269897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,power_law_1.01,0.09715200066566468
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,balanced,0.3710346619288127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,power_law_1.01,0.17073919773101806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,power_law_1.01,0.10001280307769775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,power_law_1.01,0.21507198810577394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,power_law_1.01,0.10609920024871826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,power_law_1.01,0.11549439430236816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,power_law_1.01,0.25842559337615967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,power_law_1.01,0.12110719680786133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,power_law_1.01,0.3401535987854004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,power_law_1.01,0.046489599347114566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,power_law_1.01,0.13428479433059692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,power_law_1.01,0.41249918937683105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,power_law_1.01,0.055193597078323366
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,power_law_1.01,0.15174399614334105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,power_law_1.01,0.057145601511001586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,power_law_1.01,0.1716928005218506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,power_law_1.01,0.5822783946990967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,power_law_1.01,0.05796480178833008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,power_law_1.01,0.21749119758605956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,power_law_1.01,0.751532793045044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,power_law_1.01,0.100108802318573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,power_law_1.01,0.26270079612731934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,power_law_1.01,0.10205440521240235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,power_law_1.01,0.3523008108139038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,power_law_1.01,1.0720255851745606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,power_law_1.01,0.10647679567337036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,power_law_1.01,0.4324927806854248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,power_law_1.01,0.10855040550231934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,power_law_1.01,0.02423679977655411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,power_law_1.01,0.11276160478591919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,power_law_1.01,0.5968448162078858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,power_law_1.01,0.024172799289226533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,power_law_1.01,0.1209663987159729
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,power_law_1.01,0.025446400046348572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,power_law_1.01,0.13093760013580322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,power_law_1.01,0.7708415985107422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,64,power_law_1.2,0.07772160172462464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,power_law_1.01,0.13891839981079102
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,power_law_1.01,0.025388801097869874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,64,power_law_1.2,0.08247680068016053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,power_law_1.01,0.1594879984855652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,power_law_1.01,1.1126144409179688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,power_law_1.01,0.19560320377349855
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,64,power_law_1.2,0.09508479833602905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,power_law_1.01,0.22699520587921143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,64,power_law_1.2,0.1019968032836914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,power_law_1.01,1.4292223930358887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,power_law_1.01,0.2989887952804565
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,64,power_law_1.2,0.12709759473800658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,power_law_1.2,0.04599039852619171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,power_law_1.01,1.382374382019043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,power_law_1.01,0.3729536056518555
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,64,power_law_1.2,0.147161602973938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,power_law_1.2,0.03392640054225922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,64,power_law_1.2,0.1990399956703186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,power_law_1.01,0.5179391860961914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,power_law_1.2,0.03443840146064758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,power_law_1.2,0.0348800003528595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,64,power_law_1.2,0.2634880065917969
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,power_law_1.01,0.6573823928833008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,power_law_1.2,0.03559040129184723
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,64,power_law_1.2,0.37468159198760986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,power_law_1.2,0.03681919872760773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,power_law_1.01,0.938969612121582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,power_law_1.2,0.041331198811531064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,64,power_law_1.2,0.4786687850952148
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,power_law_1.2,0.047014400362968445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,power_law_1.01,1.2290559768676759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,power_law_1.2,0.04585599899291992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,power_law_1.2,0.04757120013237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,power_law_1.2,0.04619520008563995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,power_law_1.2,0.04784640073776245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,power_law_1.2,0.04856959879398346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,power_law_1.2,0.04392960071563721
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,power_law_1.01,1.7877248764038085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,power_law_1.01,0.0262719988822937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,power_law_1.2,0.09120640158653259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,power_law_1.2,0.04613119959831238
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,power_law_1.01,0.029120001196861266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,power_law_1.2,0.09774720072746276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,power_law_1.2,0.051520001888275144
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,power_law_1.01,0.0321727991104126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,power_law_1.2,0.10071040391921997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,power_law_1.2,0.10856959819793702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,power_law_1.2,0.055206400156021115
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,power_law_1.01,0.03375360071659088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,power_law_1.01,2.327462387084961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,power_law_1.2,0.11723519563674926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,power_law_1.2,0.05581439733505249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,power_law_1.01,0.0406464010477066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,power_law_1.2,0.045184001326560974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,power_law_1.2,0.12206720113754273
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,64,power_law_1.2,0.7380608081817627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,power_law_1.2,0.0679040014743805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,power_law_1.01,0.04078719913959503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,power_law_1.2,0.04528000056743622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,power_law_1.2,0.0702015995979309
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,power_law_1.01,0.06291199922561645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,power_law_1.2,0.04552960097789764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,64,power_law_1.2,0.9493439674377442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,power_law_1.2,0.09136639833450318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,power_law_1.01,0.05301759839057922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,power_law_1.2,0.04504320025444031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,power_law_1.2,0.09784960150718688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,power_law_1.01,0.046758401393890384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,power_law_1.2,0.04674560129642487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,power_law_1.2,0.10120320320129395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,power_law_1.01,0.04268800020217896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,power_law_1.2,0.05416319966316223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,power_law_1.2,0.10619519948959351
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,power_law_1.01,0.04279040098190308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,power_law_1.2,0.11290880441665649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,power_law_1.2,0.056992000341415404
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,power_law_1.01,0.04447999894618988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,power_law_1.2,0.12076159715652465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,power_law_1.2,0.05538560152053833
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,power_law_1.01,0.04390400052070618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,power_law_1.2,0.13459839820861816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,power_law_1.2,0.13411840200424194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,power_law_1.01,0.04682239890098572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,power_law_1.2,0.09926400184631348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,power_law_1.2,0.15392639636993408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,power_law_1.2,0.15389440059661866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,power_law_1.01,0.05248000025749207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,power_law_1.2,0.1760256052017212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,power_law_1.2,0.10003839731216431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,power_law_1.2,0.18016639947891236
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,power_law_1.01,0.06318719983100891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,power_law_1.2,0.2205120086669922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,power_law_1.2,0.10631680488586426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,power_law_1.2,0.22719359397888184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,power_law_1.01,0.0763584017753601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,power_law_1.2,0.2679935932159424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,power_law_1.2,0.26935040950775146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,power_law_1.01,0.10376319885253907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,power_law_1.2,0.3570240020751953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,power_law_1.2,0.3595776081085205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,power_law_1.01,0.12796159982681274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,power_law_1.2,0.44205441474914553
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,power_law_1.01,0.17710720300674437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,power_law_1.2,0.4495039939880371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,power_law_1.2,0.6127232074737549
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,power_law_1.01,0.23377280235290526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,power_law_1.2,0.6138815879821777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,64,power_law_1.2,0.5995840072631836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,power_law_1.01,0.3323647975921631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,power_law_1.2,0.8161472320556641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,power_law_1.2,0.8038016319274902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,power_law_1.01,0.42237439155578616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,power_law_1.2,0.10852479934692383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,power_law_1.2,1.113804817199707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,power_law_1.2,0.11438080072402954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,power_law_1.2,1.1511103630065918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,power_law_1.2,0.12337280511856079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,power_law_1.2,0.13241599798202514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,power_law_1.2,1.5264575958251954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,power_law_1.2,0.14172159433364867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,power_law_1.2,1.508467197418213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,64,balanced,0.03142400085926056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,64,balanced,0.033200000723203026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,power_law_1.2,0.16019840240478517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,64,balanced,0.03317866722742716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,64,balanced,0.0376800000667572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,64,balanced,0.03873066604137421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,64,balanced,0.039749334255854286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,power_law_1.2,0.19882880449295043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,64,balanced,0.04529599845409393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,64,balanced,0.046757335464159645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,64,balanced,0.047839999198913574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,64,balanced,0.04860800007979075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,64,balanced,0.0580213318268458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,power_law_1.2,0.2353663921356201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,64,balanced,0.062234664956728615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,64,balanced,0.06427200138568878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,64,balanced,0.06834133466084798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,power_law_1.2,0.31896319389343264
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,power_law_1.2,0.026022401452064515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,power_law_1.2,0.3892352104187012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,power_law_1.2,0.025414401292800905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,power_law_1.2,0.5378623962402344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,power_law_1.2,0.026815998554229736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,power_law_1.2,0.7143104076385498
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,power_law_1.2,0.026867198944091796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,power_law_1.2,0.02736000120639801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,power_law_1.2,1.0163711547851562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,power_law_1.2,0.030739200115203858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,power_law_1.2,0.032627201080322264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,power_law_1.2,1.3386495590209961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,power_law_1.2,0.033907198905944826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,power_law_1.2,0.04170880019664765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,power_law_1.2,0.04185599982738495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,power_law_1.2,0.05902720093727112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,power_law_1.2,0.05106559991836548
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,power_law_1.2,0.04756479859352112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,power_law_1.2,0.042208001017570496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,power_law_1.2,0.04261119961738587
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,power_law_1.2,0.043782401084899905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,power_law_1.2,0.046675199270248414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,power_law_1.2,0.05119360089302063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,64,balanced,0.029616000751654308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,64,balanced,0.03142400085926056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,power_law_1.2,0.055667197704315184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,64,balanced,0.0329120010137558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,64,balanced,0.03500800083080927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,64,balanced,0.03551466763019562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,64,balanced,0.037765334049860634
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,power_law_1.2,0.06720640063285828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,64,balanced,0.043925335009892784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,64,balanced,0.04399999976158142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,64,balanced,0.04561600089073181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,64,power_law_1.01,0.034457600116729735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,64,balanced,0.04763199885686239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,power_law_1.2,0.08134400248527526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,64,balanced,0.062447999914487205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,64,balanced,0.0643039991458257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,64,power_law_1.01,0.035872000455856326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,64,balanced,0.068122665087382
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,power_law_1.2,0.10832639932632446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,64,power_law_1.01,0.03250559866428375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,64,power_law_1.01,0.0357120007276535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,64,power_law_1.01,0.033452799916267394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,64,power_law_1.01,0.0353983998298645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,power_law_1.2,1.920947265625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,64,power_law_1.01,0.033164799213409424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,64,power_law_1.01,0.03239679932594299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,64,power_law_1.01,0.03451519906520843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,64,power_law_1.01,0.03617280125617981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,64,power_law_1.01,0.04182400107383728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,power_law_1.2,2.6744192123413084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,64,power_law_1.01,0.042828801274299624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,64,power_law_1.2,0.03643519878387451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,64,power_law_1.01,0.04542720019817352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,64,balanced,0.07247466842333476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,64,power_law_1.2,0.03537920117378235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,64,power_law_1.01,0.04591360092163086
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,power_law_1.2,0.13882240056991577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,64,power_law_1.01,0.060140800476074216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,64,power_law_1.01,0.035359999537467955
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,power_law_1.2,0.19214080572128295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,64,power_law_1.01,0.06394240260124207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,64,power_law_1.01,0.03775359988212586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,64,power_law_1.01,0.06585599780082703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,64,power_law_1.01,0.04439679980278015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,64,power_law_1.01,0.07219840288162231
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,64,power_law_1.01,0.04533120095729828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,64,power_law_1.01,0.050809597969055174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,64,power_law_1.01,0.050361597537994386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,64,balanced,0.07558933397134145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,64,balanced,0.07943999767303467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,64,power_law_1.01,0.057107198238372806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,64,balanced,0.08918933073679607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,64,balanced,0.1048959990342458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,64,power_law_1.01,0.060736000537872314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,64,balanced,0.11718400319417317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,64,balanced,0.1448799967765808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,64,power_law_1.01,0.061945599317550656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,64,power_law_1.2,0.03341439962387085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,64,balanced,0.17472533384958902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,64,balanced,0.24860266844431558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,64,power_law_1.2,0.034969601035118106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,64,balanced,0.3062666654586792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,64,power_law_1.2,0.035750401020050046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,64,balanced,0.4301226536432902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,64,balanced,0.543994665145874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,power_law_1.2,0.249017596244812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,64,balanced,0.7893493175506592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,power_law_1.2,0.3553663969039917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,power_law_1.2,0.4757631778717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,64,power_law_1.2,0.033395200967788696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,64,power_law_1.01,0.07155200242996215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,64,power_law_1.2,0.03340800106525421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,64,power_law_1.2,1.9545536041259766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,64,power_law_1.2,0.03348479866981506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,64,balanced,0.041450666884581246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,64,balanced,0.05644799768924713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,64,power_law_1.2,0.03358719944953918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,64,balanced,0.06460266808668773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,64,balanced,0.11013333002726237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,64,balanced,0.1458293298880259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,64,power_law_1.2,0.03437440097332001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,64,balanced,0.1722453236579895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,64,balanced,0.1674720048904419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,64,power_law_1.2,0.03633280098438263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,64,balanced,0.17173333962758383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,64,balanced,0.14627200365066528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,64,balanced,1.0303093592325847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,64,balanced,0.14990400274594626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,64,balanced,0.037690666814645134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,64,balanced,0.043061330914497375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,64,balanced,0.2911786635716756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,64,balanced,0.054085334142049156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,64,balanced,0.0746506651242574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,64,balanced,0.09079466263453166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,64,balanced,0.10724799831708272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,64,balanced,0.12369599938392639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,64,balanced,0.12921599547068277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,64,balanced,0.11958932876586914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,64,balanced,0.12202133735020955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,64,balanced,0.2804693380991618
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,64,power_law_1.01,0.07882879972457886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,64,balanced,0.29955732822418213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,64,power_law_1.2,0.038412800431251524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,64,power_law_1.01,0.08277119994163513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,64,power_law_1.2,0.04533120095729828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,64,power_law_1.01,0.09240959882736206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,64,power_law_1.2,0.04524799883365631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,64,power_law_1.01,0.11031039953231811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,64,power_law_1.2,0.050220799446105954
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,64,power_law_1.01,0.1297919988632202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,64,power_law_1.2,0.042028799653053284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,64,power_law_1.2,0.05255680084228516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,64,power_law_1.01,0.1647104024887085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,64,balanced,0.311845342318217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,64,power_law_1.2,0.05808640122413635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,64,balanced,0.32900800307591754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,64,power_law_1.01,0.20245120525360108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,64,balanced,0.3468746741612752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,64,balanced,0.3678133487701416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,64,power_law_1.01,0.28482561111450194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,64,balanced,0.3918773333231608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,64,balanced,0.4349173307418823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,64,power_law_1.01,0.3599488019943237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,64,balanced,0.5236320098241171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,64,balanced,0.31481067339579266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,64,balanced,0.5959733327229818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,64,balanced,0.33872000376383465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,64,power_law_1.01,0.5234111785888672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,64,balanced,0.36314666271209717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,64,power_law_1.01,0.0763584017753601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,64,balanced,0.7335039774576823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,64,balanced,0.38950932025909424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,64,power_law_1.01,0.08076800107955932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,64,power_law_1.01,0.6588736057281495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,64,balanced,0.4389280080795288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,64,balanced,0.07876266539096832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,64,power_law_1.01,0.09231359958648681
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,64,balanced,0.08475733796755473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,64,balanced,0.5144213438034058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,64,power_law_1.2,0.0431551992893219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,64,balanced,0.0920799970626831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,64,power_law_1.01,0.10638079643249512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,64,balanced,0.1097866694132487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,64,power_law_1.2,0.0457152009010315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,64,balanced,0.12574932972590128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,64,power_law_1.01,0.12455040216445923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,64,power_law_1.2,0.061849600076675414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,64,balanced,0.1585546632607778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,64,power_law_1.2,0.04584319889545441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,64,balanced,0.19301867485046387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,64,power_law_1.2,0.06335999965667724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,64,power_law_1.2,0.06047359704971313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,64,power_law_1.2,0.07217280268669128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,64,power_law_1.2,0.06415359973907471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,64,power_law_1.2,0.06689919829368592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,64,power_law_1.2,0.07304319739341736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,64,balanced,0.8723733425140381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,64,balanced,1.2083626588185628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,64,balanced,0.5949920018513998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,64,power_law_1.01,0.9579648017883301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,64,balanced,0.7508052984873453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,64,balanced,1.4899253845214844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,64,power_law_1.01,0.155622398853302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,64,balanced,0.2717120051383972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,64,balanced,0.9081546465555826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,64,power_law_1.01,0.19839359521865846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,64,balanced,0.3402293523152669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,64,power_law_1.01,1.2410880088806153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,64,balanced,0.05212800204753876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,64,power_law_1.01,0.2764863967895508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,64,balanced,2.1090826988220215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,64,balanced,0.06069866816202799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,64,balanced,1.2526240348815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,64,power_law_1.01,0.3441663980484009
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,64,power_law_1.01,0.4900991916656494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,64,balanced,1.5708640416463215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,64,balanced,2.6748479207356772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,64,power_law_1.01,0.6425216197967529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,64,balanced,2.2332480748494468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,64,power_law_1.01,0.9263423919677735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,64,balanced,0.4793066581090291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,64,balanced,3.959381421407064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,64,balanced,0.07014399766921997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,64,power_law_1.01,1.2377728462219237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,64,balanced,2.867738723754883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,64,balanced,4.234106699625651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,64,balanced,5.159626642862956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,64,power_law_1.2,0.07861760258674622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,64,power_law_1.2,0.0838591992855072
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,64,power_law_1.01,0.07884160280227662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,64,power_law_1.2,0.09622399806976319
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,64,balanced,0.6107413371404012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,64,power_law_1.2,0.11172480583190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,64,power_law_1.01,0.07791360020637512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,64,balanced,0.8852533499399821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,64,power_law_1.2,0.13331199884414674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,64,power_law_1.01,0.0791104018688202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,64,balanced,1.1541546980539958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,64,balanced,5.545642852783203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,64,power_law_1.01,0.08074880242347718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,64,power_law_1.01,0.08812159895896912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,64,power_law_1.2,0.11816320419311524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,64,power_law_1.01,0.07970560193061829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,64,power_law_1.2,0.11755520105361938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,64,power_law_1.01,0.08729599714279175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,64,power_law_1.01,0.09903360009193421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,64,balanced,0.08372799555460612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,64,power_law_1.01,0.1195904016494751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,64,power_law_1.2,0.11913599967956542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,64,power_law_1.01,0.08739839792251587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,64,power_law_1.01,0.1153216004371643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,64,power_law_1.2,0.12314239740371705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,64,power_law_1.01,0.11892479658126831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,64,power_law_1.01,0.1226304054260254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,64,power_law_1.01,0.09100800156593322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,64,power_law_1.2,0.17273600101470948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,64,power_law_1.01,0.11249920129776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,64,power_law_1.2,0.21242239475250244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,64,power_law_1.2,0.13642879724502563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,64,power_law_1.01,0.08641279935836792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,64,power_law_1.01,0.11806720495223999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,64,power_law_1.2,0.3022336006164551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,64,power_law_1.2,0.1687999963760376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,64,power_law_1.01,0.10043519735336304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,64,power_law_1.01,0.2729919910430908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,64,power_law_1.2,0.38163840770721436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,64,power_law_1.2,3.310617446899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,64,power_law_1.01,0.1081663966178894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,64,power_law_1.2,0.16264959573745727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,64,power_law_1.2,0.07628160119056701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,64,power_law_1.01,0.2949568033218384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,64,balanced,0.0939573347568512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,64,power_law_1.2,0.5469759941101074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,64,power_law_1.2,0.08329600095748901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,64,balanced,0.10414399703343709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,64,power_law_1.2,0.07686399817466735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,64,power_law_1.01,0.11471359729766846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,64,power_law_1.01,0.30976641178131104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,64,power_law_1.2,0.0932096004486084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,64,power_law_1.2,0.7030335903167725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,64,power_law_1.2,0.11083519458770752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,64,power_law_1.01,0.30969600677490233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,64,power_law_1.2,0.07655680179595947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,64,power_law_1.01,0.3349760055541992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,64,power_law_1.2,0.13148159980773927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,64,power_law_1.01,0.12104959487915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,64,power_law_1.01,0.32371840476989744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,64,power_law_1.2,0.0778880000114441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,64,power_law_1.01,0.122489595413208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,64,power_law_1.2,0.08087040185928344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,64,power_law_1.01,0.13639039993286134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,64,power_law_1.2,0.0847487986087799
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,64,power_law_1.01,0.16591999530792237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,64,balanced,0.11026133100191753
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,64,balanced,0.11446932951609294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,64,balanced,0.3165439963340759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,64,power_law_1.01,0.154367995262146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,64,balanced,0.3264266649881999
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,64,balanced,0.340831995010376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,64,power_law_1.01,0.16277120113372803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,64,balanced,0.3585439920425415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,64,power_law_1.01,0.3658560037612915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,64,balanced,0.3758346637090047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,64,power_law_1.01,0.13153280019760133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,64,balanced,0.4076319932937622
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,64,power_law_1.01,0.4005760192871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,64,power_law_1.01,0.3319808006286621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,64,balanced,0.4374986489613851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,64,power_law_1.01,0.144268798828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,64,balanced,0.4696799914042155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,64,power_law_1.01,0.4553215980529785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,64,balanced,0.5332320133845011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,64,power_law_1.01,0.2852288007736206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,64,balanced,0.654581348101298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,64,power_law_1.01,0.5184192180633544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,64,power_law_1.01,0.3080575942993164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,64,power_law_1.2,0.16725120544433594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,64,balanced,0.7790506680806478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,64,power_law_1.2,0.10073599815368653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,64,power_law_1.01,0.6050367832183838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,64,power_law_1.2,0.14086400270462035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,64,balanced,1.0788586934407551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,64,power_law_1.2,0.14193919897079468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,64,power_law_1.2,0.11921279430389405
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,64,power_law_1.01,0.7750400066375732
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,64,balanced,1.3238933086395264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,64,power_law_1.2,0.17375999689102173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,64,power_law_1.2,0.28538880348205564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,64,power_law_1.2,0.12407679557800293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,64,power_law_1.2,0.20620160102844237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,64,power_law_1.2,0.30303359031677246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,64,power_law_1.01,0.9421440124511719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,64,balanced,1.8709972699483235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,64,power_law_1.2,0.11117440462112427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,64,power_law_1.2,0.3023231983184814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,64,power_law_1.01,0.3540287971496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,64,power_law_1.2,0.11658240556716919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,64,power_law_1.2,0.38017280101776124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,64,power_law_1.01,0.3710655927658081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,64,power_law_1.01,1.2959360122680663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,64,power_law_1.2,0.27548799514770506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,64,balanced,2.362629254659017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,64,power_law_1.2,0.5452415943145752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,64,power_law_1.01,0.4013887882232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,64,power_law_1.01,0.32327680587768554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,64,power_law_1.2,0.29412479400634767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,64,power_law_1.01,0.34773759841918944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,64,power_law_1.01,0.43984642028808596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,64,power_law_1.2,0.3117568016052246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,64,power_law_1.01,1.6298303604125977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,64,power_law_1.01,0.3790015935897827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,64,power_law_1.01,0.4791232109069824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,64,power_law_1.2,0.346617603302002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,64,balanced,3.453674634297689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,64,power_law_1.01,0.40930562019348143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,64,power_law_1.01,0.5489535808563233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,64,power_law_1.2,0.325926399230957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,64,power_law_1.2,0.36981759071350095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,64,power_law_1.01,0.46976637840270996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,64,power_law_1.2,0.36136960983276367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,64,power_law_1.01,0.697817611694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,64,power_law_1.01,0.5308095932006835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,64,power_law_1.2,0.4071360111236572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,64,power_law_1.01,0.6126527786254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,64,power_law_1.2,0.39466879367828367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,64,balanced,4.49124272664388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,64,power_law_1.01,0.8525504112243653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,64,power_law_1.2,0.45981440544128416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,64,power_law_1.2,0.7020864009857177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,64,power_law_1.2,0.42851839065551756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,64,power_law_1.2,0.5231679916381836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,64,power_law_1.01,1.1484864234924317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,64,power_law_1.2,0.47867522239685056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,64,power_law_1.2,0.6141439914703369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,64,power_law_1.01,1.4683391571044921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,64,power_law_1.2,0.5506752014160157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,64,power_law_1.01,2.3580352783203127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,64,balanced,6.6421864827473955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,64,power_law_1.2,0.6500927925109863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,64,power_law_1.2,1.055072021484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,64,power_law_1.01,2.0451135635375977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,64,power_law_1.01,3.0347904205322265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,64,power_law_1.2,1.3921664237976075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,64,power_law_1.01,2.6425407409667967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,64,balanced,8.72098159790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,128,balanced,0.05667200187842051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,128,balanced,0.05056533217430115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,128,balanced,0.052000001072883606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,128,balanced,0.05151999990145365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,64,power_law_1.01,4.450201416015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,128,balanced,0.057989334066708885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,128,balanced,0.08801600337028503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,128,balanced,0.09598933657010396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,128,balanced,0.09702400366465251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,128,balanced,0.11432533462842305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,64,power_law_1.01,3.858687973022461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,128,balanced,0.11761599779129028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,128,balanced,0.17268800735473633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,128,balanced,0.17908267180124918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,64,power_law_1.01,0.7869184017181396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,128,balanced,0.18330667416254678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,128,balanced,0.19022399187088013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,64,power_law_1.01,0.9628800392150879
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,64,power_law_1.01,1.2876480102539063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,64,power_law_1.01,5.08012809753418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,64,power_law_1.01,5.813811111450195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,128,balanced,0.04186666508515676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,128,balanced,0.041589332123597465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,128,balanced,0.04155199974775314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,64,power_law_1.01,1.6309696197509767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,64,power_law_1.2,0.7882880210876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,64,power_law_1.2,0.9736767768859863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,64,power_law_1.2,0.8063936233520508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,64,power_law_1.2,1.0710080146789551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,64,power_law_1.01,2.262054443359375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,64,power_law_1.2,0.08626559972763062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,64,power_law_1.2,0.982636833190918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,64,power_law_1.2,0.08654080033302307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,128,balanced,0.04214400053024292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,128,balanced,0.04359999795754751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,64,power_law_1.2,1.3568320274353027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,64,power_law_1.2,0.08711680173873901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,64,power_law_1.01,7.37805404663086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,64,power_law_1.01,3.085804748535156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,64,power_law_1.2,1.727065658569336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,64,power_law_1.2,1.3145855903625487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,64,power_law_1.2,2.441555213928223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,64,power_law_1.01,4.446828842163086
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,128,balanced,0.06434133152167003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,128,balanced,0.08069866895675659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,64,power_law_1.2,1.6952896118164062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,128,balanced,0.08351999521255493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,128,balanced,0.08443199594815572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,128,balanced,0.08694932858149211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,128,balanced,0.15586666266123453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,128,balanced,0.16676799456278482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,128,balanced,0.1726293365160624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,128,balanced,0.1808746655782064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,128,balanced,0.1954453388849894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,128,balanced,0.20699199040730795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,128,balanced,0.22612265745798746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,64,power_law_1.2,3.3015487670898436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,128,balanced,0.25834665695826214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,128,balanced,0.2905866702397664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,128,balanced,0.35918935139973956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,64,power_law_1.01,9.84686050415039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,128,balanced,0.42829867204030353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,128,balanced,0.058575997749964394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,128,balanced,0.06004266440868378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,128,balanced,0.05792533357938131
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,128,balanced,0.059994667768478394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,128,balanced,0.06029333174228668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,64,power_law_1.01,5.932870483398437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,128,balanced,0.08726400136947632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,128,balanced,0.08686932921409607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,128,balanced,0.09203733007113139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,128,balanced,0.18083733320236206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,128,balanced,0.18517333269119263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,128,balanced,0.19131733973821005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,128,balanced,0.19706666469573975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,128,balanced,0.2019360065460205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,128,balanced,0.21374932924906412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,64,power_law_1.2,2.4290815353393556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,128,balanced,0.22522666056950888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,128,balanced,0.23775466283162436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,64,power_law_1.2,4.91338882446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,128,balanced,0.26049067576726276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,64,power_law_1.2,0.0902079999446869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,128,balanced,0.308351993560791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,128,balanced,0.3596160014470418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,128,balanced,0.5607626835505167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,128,balanced,0.4586453437805176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,128,balanced,0.6972320079803467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,128,balanced,0.5595626831054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,128,balanced,0.9772799809773763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,64,power_law_1.2,3.1099456787109374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,128,balanced,0.7733279863993326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,128,balanced,1.2475519975026448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,128,balanced,0.9724213282267252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,128,balanced,1.3900853792826335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,128,balanced,1.7958186467488606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,128,balanced,0.20258132616678873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,128,balanced,1.8061760266621907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,128,balanced,0.21189866463343301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,64,power_law_1.2,0.08462079763412475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,128,balanced,2.345344066619873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,128,balanced,0.23017599185307822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,128,balanced,0.2603519956270854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,128,power_law_1.01,0.06383360028266907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,64,power_law_1.2,4.697196960449219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,128,balanced,2.6382506688435874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,128,power_law_1.01,0.0627135992050171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,128,power_law_1.01,0.06319360136985779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,128,power_law_1.01,0.06362239718437195
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,128,balanced,0.027503999571005504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,128,power_law_1.01,0.06999679803848266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,128,balanced,3.465754508972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,128,balanced,0.02828799933195114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,128,balanced,0.029279999434947968
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,128,balanced,0.029616000751654308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,128,power_law_1.01,0.08541439771652222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,128,balanced,0.033344000577926636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,128,power_law_1.01,0.047660800814628604
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,128,balanced,0.03669866671164831
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,128,balanced,0.08327466746171315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,128,power_law_1.01,0.09457280039787293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,128,balanced,0.0535093347231547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,128,power_law_1.01,0.0488319993019104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,128,balanced,0.08898133039474487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,128,balanced,0.29016000032424927
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,128,balanced,0.08690133690834045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,128,power_law_1.01,0.09520000219345093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,128,balanced,0.07027733325958252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,128,balanced,0.3473066488901774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,128,power_law_1.01,0.04893440008163452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,128,balanced,0.40822935104370117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,128,power_law_1.01,0.1134335994720459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,128,balanced,0.5310399929682413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,128,power_law_1.01,0.049619200825691226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,128,power_law_1.01,0.11592320203781128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,128,balanced,0.6539626518885294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,128,power_law_1.01,0.05196160078048706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,128,power_law_1.01,0.16981120109558107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,128,balanced,0.9183253447214762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,128,power_law_1.01,0.06275839805603027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,128,power_law_1.01,0.17777279615402222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,128,balanced,1.163925329844157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,128,power_law_1.01,0.07816320061683654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,128,power_law_1.01,0.1823040008544922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,128,power_law_1.01,0.1901568055152893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,128,power_law_1.01,0.07954559922218322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,128,balanced,1.6734827359517415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,128,power_law_1.01,0.2015552043914795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,128,power_law_1.01,0.08184959888458251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,128,balanced,0.05412266651789347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,128,power_law_1.01,0.2109760046005249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,128,balanced,0.05961066484451294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,128,power_law_1.01,0.08474239706993103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,128,balanced,0.05663999915122986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,128,balanced,0.05175999800364176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,128,balanced,2.185002644856771
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,128,balanced,0.05383466680844625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,128,power_law_1.01,0.1537343978881836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,128,balanced,0.0574186642964681
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,128,balanced,0.065461332599322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,128,power_law_1.01,0.16521600484848023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,128,power_law_1.01,0.07482240200042725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,64,power_law_1.2,0.10181759595870972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,128,power_law_1.01,0.17075200080871583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,128,power_law_1.01,0.07385600209236146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,128,power_law_1.01,0.17982079982757568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,64,power_law_1.2,0.10949759483337403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,128,power_law_1.01,0.07339519858360291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,128,power_law_1.01,0.19393279552459716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,64,power_law_1.2,0.11228799819946289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,128,power_law_1.01,0.07306879758834839
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,128,power_law_1.01,0.20432000160217284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,128,power_law_1.01,0.07390720248222352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,128,power_law_1.01,0.2287424087524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,128,power_law_1.01,0.08431360125541687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,128,power_law_1.01,0.23214080333709716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,128,power_law_1.01,0.26311678886413575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,64,power_law_1.2,6.230931091308594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,128,power_law_1.01,0.26396799087524414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,128,power_law_1.01,0.08882560133934021
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,128,balanced,0.07443733513355255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,128,balanced,0.09857066472371419
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,128,power_law_1.01,0.2986624002456665
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,128,balanced,0.11916800340016682
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,128,balanced,0.15820800264676413
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,128,balanced,0.20164799690246582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,128,power_law_1.01,0.3834431886672974
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,128,balanced,0.29366399844487506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,128,power_law_1.01,0.46233601570129396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,128,balanced,0.3773226737976074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,64,power_law_1.2,1.3752063751220702
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,128,balanced,0.5378559827804565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,128,power_law_1.01,0.6000383853912353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,128,power_law_1.2,0.06323199868202209
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,128,balanced,0.7015840212504069
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,128,power_law_1.01,0.7724736213684082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,128,power_law_1.2,0.06295679807662964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,128,power_law_1.2,0.04832639992237091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,128,power_law_1.01,0.2965951919555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,128,power_law_1.2,0.0632960021495819
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,128,power_law_1.2,0.048614400625228885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,128,power_law_1.01,0.09144960045814514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,128,power_law_1.01,0.3700608015060425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,128,power_law_1.01,1.083238410949707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,128,power_law_1.2,0.048851200938224794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,128,power_law_1.01,0.17985279560089112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,128,power_law_1.01,0.450105619430542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,128,power_law_1.2,0.0493120014667511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,128,power_law_1.01,0.18306560516357423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,128,power_law_1.01,1.3881152153015137
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,128,power_law_1.01,0.18920960426330566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,128,power_law_1.01,0.6054272174835205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,64,power_law_1.2,6.103116989135742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,128,power_law_1.01,0.1953727960586548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,128,power_law_1.01,0.753446388244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,128,power_law_1.01,0.20215039253234862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,128,power_law_1.2,0.0710528016090393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,128,power_law_1.01,2.0162303924560545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,128,power_law_1.01,0.21270399093627929
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,128,power_law_1.2,0.07075200080871583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,128,power_law_1.01,1.0585663795471192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,128,power_law_1.01,0.22581760883331298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,128,power_law_1.2,0.0641152024269104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,128,power_law_1.2,0.07148159742355346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,128,power_law_1.01,0.23731839656829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,128,power_law_1.2,0.06784639954566955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,128,power_law_1.2,0.07151359915733338
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,128,power_law_1.01,0.26888959407806395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,128,power_law_1.01,1.360979175567627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,128,power_law_1.2,0.08537600040435792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,128,power_law_1.2,0.05235199928283692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,128,power_law_1.2,0.07253760099411011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,128,power_law_1.01,0.32778239250183105
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,128,power_law_1.01,0.02741760015487671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,128,power_law_1.2,0.09375360012054443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,128,power_law_1.2,0.06281599998474122
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,128,power_law_1.2,0.08570240139961242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,128,power_law_1.01,0.39623680114746096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,128,power_law_1.01,0.02834559977054596
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,128,power_law_1.2,0.09497600197792053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,128,power_law_1.2,0.0775168001651764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,128,power_law_1.2,0.08832640051841736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,128,power_law_1.01,1.9471040725708009
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,128,power_law_1.01,0.028569599986076354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,128,power_law_1.01,0.527027177810669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,128,power_law_1.2,0.07957119941711426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,128,power_law_1.2,0.09057279825210571
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,128,power_law_1.01,0.030399999022483824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,128,power_law_1.2,0.08394240140914917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,128,power_law_1.01,0.6623807907104492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,128,power_law_1.2,0.1793023943901062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,128,power_law_1.2,0.08672639727592468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,128,power_law_1.2,0.18345600366592407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,128,power_law_1.2,0.1540287971496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,128,power_law_1.01,2.601158332824707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,128,power_law_1.01,2.744076728820801
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,128,power_law_1.2,0.18944640159606935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,128,power_law_1.2,0.1648959994316101
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,128,power_law_1.2,0.17078399658203125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,128,power_law_1.2,0.1957759976387024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,128,power_law_1.2,0.18044159412384034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,128,power_law_1.2,0.20181119441986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,128,power_law_1.2,0.1133504033088684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,128,power_law_1.2,0.21527040004730225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,64,power_law_1.2,0.30883839130401614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,128,power_law_1.2,0.11624959707260132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,128,power_law_1.2,0.22716159820556642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,64,power_law_1.2,0.319321608543396
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,128,power_law_1.01,0.03312639892101288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,128,power_law_1.2,0.16949119567871093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,128,power_law_1.2,0.24320640563964843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,64,power_law_1.2,0.33221120834350587
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,128,power_law_1.01,0.03528960049152374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,128,power_law_1.2,0.1770624041557312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,128,power_law_1.2,0.27311999797821046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,64,power_law_1.2,0.3507200002670288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,128,balanced,0.04364799956480662
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,128,power_law_1.01,0.08204799890518188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,128,balanced,0.04513599971930186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,128,balanced,0.04399466514587402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,128,power_law_1.2,0.3440000057220459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,128,balanced,0.04390933116277059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,64,power_law_1.2,0.37253119945526125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,128,balanced,0.04987200101216634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,128,balanced,0.07903466622034709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,128,balanced,0.08545066912968953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,128,balanced,0.08682133754094441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,128,power_law_1.2,0.4045119762420654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,128,balanced,0.10287466645240784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,64,power_law_1.2,0.4124927997589111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,128,balanced,0.10318400462468465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,128,balanced,0.15102932850519815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,128,balanced,0.15813333789507547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,128,balanced,0.16263999541600546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,128,power_law_1.2,0.5501440048217774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,128,balanced,0.16972267627716064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,128,balanced,0.17971199750900269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,128,balanced,0.18761066595713297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,128,balanced,0.2039519945780436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,128,power_law_1.2,0.6920000076293945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,128,balanced,0.23198399941126505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,128,balanced,0.26078399022420246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,128,power_law_1.01,0.913088035583496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,128,balanced,0.3166186610857646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,128,balanced,0.37111465136210126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,128,power_law_1.2,0.18215680122375488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,128,power_law_1.2,0.9904064178466797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,128,balanced,0.48124265670776367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,128,power_law_1.2,0.1894592046737671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,128,power_law_1.01,1.160812759399414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,128,balanced,0.5932533343633016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,128,power_law_1.2,0.2059391975402832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,128,balanced,0.8316213289896647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,128,power_law_1.2,1.2769791603088378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,128,power_law_1.2,0.21311359405517577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,128,balanced,1.054085334142049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,128,power_law_1.01,1.6409599304199218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,64,power_law_1.2,0.44518399238586426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,128,power_law_1.2,0.23571200370788575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,128,balanced,1.5174612998962402
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,128,power_law_1.2,0.02680320143699646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,64,power_law_1.2,0.4898560047149658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,128,power_law_1.2,1.846963119506836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,128,power_law_1.2,0.02768000066280365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,128,power_law_1.01,2.175712013244629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,128,balanced,1.9767786661783855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,64,power_law_1.2,0.5648575782775879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,128,power_law_1.2,0.03025279939174652
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,128,power_law_1.01,0.05115519762039185
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,128,power_law_1.2,0.029817599058151244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,64,power_law_1.2,0.7275263786315918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,128,power_law_1.01,0.08157439827919007
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,128,power_law_1.2,0.03290880024433136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,128,power_law_1.01,0.0815936028957367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,64,power_law_1.2,0.8815808296203613
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,128,power_law_1.2,0.03506560027599335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,128,power_law_1.01,3.1742719650268554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,128,power_law_1.01,0.0657151997089386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,128,power_law_1.2,0.26739840507507323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,128,power_law_1.01,0.052051198482513425
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,128,power_law_1.2,0.08120319843292237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,128,power_law_1.2,0.31201920509338377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,64,power_law_1.2,1.1934975624084472
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,128,power_law_1.01,0.05256320238113403
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,128,power_law_1.2,0.049670401215553286
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,128,power_law_1.01,0.05245440006256104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,128,power_law_1.2,0.39749119281768797
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,128,power_law_1.2,0.08097919821739197
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,128,power_law_1.01,0.05076479911804199
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,128,power_law_1.2,0.07932159900665284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,128,power_law_1.2,0.4790463924407959
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,128,power_law_1.2,2.4816831588745116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,128,power_law_1.01,0.051744002103805545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,128,power_law_1.01,4.23306884765625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,128,power_law_1.01,0.05715199708938599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,128,balanced,0.054117331902186074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,128,balanced,0.05411200225353241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,128,power_law_1.2,0.6598720073699951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,128,balanced,0.05608533322811127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,128,balanced,0.05624533196290334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,128,power_law_1.01,0.06762239933013917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,128,balanced,0.058090666929880776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,128,balanced,0.08169599870840709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,128,balanced,0.08156799773375194
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,128,power_law_1.01,0.08240640163421631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,128,power_law_1.2,0.8443648338317871
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,128,power_law_1.01,0.10726399421691894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,128,power_law_1.2,3.6899646759033202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,128,balanced,0.03745600084463755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,128,balanced,0.039450667798519135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,128,power_law_1.2,1.1842047691345214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,128,balanced,0.03945599993069967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,128,balanced,0.039503999054431915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,128,balanced,0.04154133299986521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,128,balanced,0.056176001826922096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,64,power_law_1.2,1.5384256362915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,128,balanced,0.0719946672519048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,128,balanced,0.07123200098673503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,128,balanced,0.07026666899522145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,128,power_law_1.2,0.06276479959487916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,128,balanced,0.07394666473070781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,128,power_law_1.2,1.555116844177246
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,128,power_law_1.2,0.05087360143661499
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,128,power_law_1.2,0.0523967981338501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,64,power_law_1.2,2.2130815505981447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,128,balanced,0.0846720039844513
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,128,power_law_1.2,0.053830397129058835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,128,balanced,0.15901866555213928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,128,balanced,0.16056000192960104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,128,balanced,0.16710933049519858
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,128,power_law_1.2,0.05134080052375793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,128,power_law_1.2,2.3747520446777344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,128,balanced,0.17489065726598105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,128,balanced,0.18201067050298056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,128,power_law_1.01,0.13009920120239257
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,128,power_law_1.2,0.051718401908874514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,128,balanced,0.1893120010693868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,128,balanced,0.20193066199620566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,128,balanced,0.21248533328374228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,128,power_law_1.01,0.18001919984817505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,128,balanced,0.23324799537658691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,64,power_law_1.2,2.785126495361328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,128,balanced,0.2837013403574626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,128,power_law_1.2,0.19530240297317505
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,128,power_law_1.01,0.23356161117553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,128,balanced,0.33058667182922363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,128,power_law_1.2,0.20597119331359864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,128,balanced,0.4238986571629842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,128,power_law_1.2,2.9811775207519533
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,128,power_law_1.01,0.3294143915176392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,128,balanced,0.5169333219528198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,128,power_law_1.2,0.23109118938446044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,128,balanced,0.7128586769104004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,128,power_law_1.2,0.26137599945068357
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,128,power_law_1.01,0.432147216796875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,128,balanced,0.897648016611735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,128,power_law_1.2,5.0276031494140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,128,power_law_1.2,0.30271360874176023
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,128,power_law_1.01,0.6136256217956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,128,power_law_1.01,0.05795199871063232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,64,power_law_1.2,4.1430400848388675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,128,power_law_1.2,0.3820544004440308
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,128,power_law_1.2,0.05799679756164551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,128,power_law_1.01,0.05800960063934326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,128,power_law_1.2,0.46364798545837405
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,128,power_law_1.2,0.07126399874687195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,128,power_law_1.01,0.058259201049804685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,128,balanced,0.14075199762980142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,128,balanced,0.1520906686782837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,128,balanced,0.15641066431999207
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,128,power_law_1.2,0.08865919709205627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,128,power_law_1.01,0.058873599767684935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,128,balanced,0.16703466574350992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,128,balanced,0.17833600441614786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,128,balanced,0.1874026656150818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,128,power_law_1.01,0.06249600052833557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,128,balanced,0.20698134104410806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,128,balanced,0.2346293330192566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,128,power_law_1.01,0.07576320171356202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,128,balanced,0.2669866681098938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,128,balanced,0.3296213348706563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,128,power_law_1.01,0.08343679904937744
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,128,balanced,0.3918986717859904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,128,balanced,1.2814453442891438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,128,balanced,0.5157226721445719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,64,power_law_1.2,5.562067031860352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,128,power_law_1.01,0.08435840010643006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,128,balanced,0.6379253466924032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,128,power_law_1.01,0.10072319507598877
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,128,power_law_1.01,0.8215999603271484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,128,balanced,0.8932533264160156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,128,power_law_1.01,0.10114560127258301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,128,power_law_1.01,0.0430400013923645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,128,power_law_1.2,0.6323904037475586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,128,power_law_1.01,0.14994560480117797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,128,balanced,1.13701335589091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,128,power_law_1.01,0.04264959990978241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,128,power_law_1.01,0.15481599569320678
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,128,power_law_1.2,0.11273599863052368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,128,power_law_1.2,0.7821951866149902
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,128,power_law_1.01,0.04352000057697296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,128,balanced,1.6400693257649739
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,128,power_law_1.2,0.13961600065231322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,128,power_law_1.01,0.04351359903812409
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,128,power_law_1.2,0.1914944052696228
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,128,balanced,0.029071999092896778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,128,power_law_1.2,1.1146304130554199
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,128,balanced,0.029445332785447437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,128,power_law_1.01,0.04636160135269165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,128,balanced,2.140223979949951
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,128,power_law_1.2,0.2544895887374878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,128,power_law_1.01,0.05496960282325745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,128,power_law_1.01,0.06304640173912049
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,128,balanced,1.6656479835510254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,128,power_law_1.01,0.06761599779129028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,128,power_law_1.2,1.448908805847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,128,power_law_1.2,0.36929919719696047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,128,power_law_1.01,0.062035202980041504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,128,power_law_1.01,0.06922240257263183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,128,power_law_1.01,0.06227200031280518
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,128,power_law_1.2,0.48380160331726074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,128,power_law_1.01,0.07049599885940552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,128,balanced,2.4310666720072427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,128,power_law_1.01,0.06347519755363465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,128,power_law_1.01,0.07320320010185241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,128,power_law_1.2,2.12737922668457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,128,power_law_1.01,0.06556159853935242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,128,power_law_1.2,0.7070655822753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,128,power_law_1.01,0.07862399816513062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,128,balanced,0.029685333371162415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,128,balanced,3.1948318481445312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,128,balanced,0.03162666658560435
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,128,balanced,0.03348266581694285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,128,power_law_1.01,0.08145279884338379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,128,balanced,0.03616533428430557
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,128,balanced,0.04785066843032837
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,128,balanced,0.07751999795436859
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,128,balanced,0.07694933315118153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,128,power_law_1.2,2.91397762298584
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,128,balanced,0.07904000083605449
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,128,balanced,0.06222933530807495
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,128,balanced,0.05007466673851013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,128,power_law_1.2,0.05862399935722351
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,128,balanced,0.052815998593966164
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,128,balanced,0.05239466826121012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,128,power_law_1.2,0.05851519703865051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,128,power_law_1.01,0.1588863968849182
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,128,balanced,0.054245332876841225
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,128,balanced,0.05129600067933401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,128,power_law_1.01,0.14095360040664673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,128,power_law_1.01,0.16672639846801757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,128,power_law_1.01,0.15207680463790893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,128,power_law_1.01,0.1768895983695984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,128,power_law_1.01,0.15637760162353515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,128,power_law_1.01,0.18624639511108398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,128,power_law_1.2,0.951756763458252
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,128,power_law_1.01,0.16491520404815674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,128,power_law_1.01,0.2079103946685791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,128,power_law_1.01,0.0839680016040802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,128,power_law_1.2,0.04245119988918304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,128,power_law_1.01,0.17824000120162964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,128,power_law_1.01,0.23578879833221436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,128,power_law_1.01,0.15645439624786378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,128,power_law_1.01,0.18865920305252076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,64,power_law_1.2,8.006515502929688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,128,power_law_1.01,0.27175679206848147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,128,power_law_1.01,0.159552001953125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,128,power_law_1.01,0.20856959819793702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,128,power_law_1.01,0.342303991317749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,128,power_law_1.01,0.1660223960876465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,128,power_law_1.01,0.23430399894714354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,128,power_law_1.2,0.05869439840316772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,128,power_law_1.01,0.40879998207092283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,128,power_law_1.01,0.17264000177383423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,128,power_law_1.01,0.2722496032714844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,128,power_law_1.2,0.058950400352478026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,128,power_law_1.01,0.17827199697494506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,128,power_law_1.01,0.5343552112579346
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,128,power_law_1.01,0.34163200855255127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,128,power_law_1.2,0.06309760212898255
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,128,power_law_1.01,0.18844159841537475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,128,power_law_1.01,0.4128448009490967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,128,power_law_1.2,0.07587199807167053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,128,power_law_1.01,0.6703680038452149
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,128,power_law_1.01,0.20079360008239747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,128,power_law_1.2,0.08316799998283386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,128,power_law_1.01,0.5473408222198486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,128,power_law_1.2,0.04361599981784821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,128,power_law_1.2,0.0847104012966156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,128,power_law_1.2,0.043558400869369504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,128,power_law_1.01,0.026956799626350402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,128,power_law_1.01,0.6797887802124023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,128,power_law_1.2,0.10037120580673217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,128,power_law_1.2,0.04487679898738861
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,128,balanced,0.05406400064627329
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,128,balanced,0.06020266811052958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,128,power_law_1.2,0.04710400104522705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,128,balanced,0.06649066507816315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,128,power_law_1.01,0.9560447692871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,128,balanced,0.08717866738637288
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,128,balanced,0.10739200313886006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,128,power_law_1.2,0.05655679702758789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,128,balanced,0.1421173314253489
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,128,balanced,0.1794453263282776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,128,power_law_1.2,0.06892160177230836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,128,power_law_1.01,1.2342144012451173
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,128,balanced,0.2619733413060506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,128,power_law_1.2,0.07133439779281617
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,128,balanced,0.3356959819793701
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,128,power_law_1.01,0.21271040439605712
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,128,power_law_1.01,0.9613247871398926
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,128,balanced,0.47516798973083496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,128,power_law_1.2,0.07014399766921997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,128,power_law_1.01,0.23980159759521485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,128,balanced,0.6174453496932983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,128,power_law_1.01,1.8025344848632812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,128,power_law_1.01,0.3036672115325928
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,128,power_law_1.01,0.02746239900588989
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,128,power_law_1.01,1.2279040336608886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,128,power_law_1.2,0.060127997398376466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,128,power_law_1.01,0.3585088014602661
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,128,power_law_1.01,0.02824319899082184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,128,power_law_1.2,0.0587007999420166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,128,power_law_1.01,0.47220478057861326
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,128,power_law_1.01,0.029600000381469725
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,128,power_law_1.2,0.058982402086257935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,128,power_law_1.01,1.8139328002929687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,128,power_law_1.01,0.5991424083709717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,128,power_law_1.2,0.06209279894828797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,128,power_law_1.2,0.06368640065193176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,128,power_law_1.01,0.8354751586914062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,128,power_law_1.2,0.07852799892425537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,128,power_law_1.2,0.07338240146636962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,128,power_law_1.01,2.301139259338379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,128,power_law_1.01,1.063532829284668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,128,power_law_1.2,0.1017024040222168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,128,power_law_1.2,0.14083839654922486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,128,power_law_1.2,0.08213760256767273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,128,power_law_1.2,0.14869760274887084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,128,power_law_1.2,0.15127040147781373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,128,power_law_1.2,0.08134400248527526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,128,power_law_1.2,0.155731201171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,128,power_law_1.2,0.15607039928436278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,128,power_law_1.01,1.5071935653686523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,128,power_law_1.2,0.15500160455703735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,128,power_law_1.2,0.1595199942588806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,128,power_law_1.2,0.16540160179138183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,128,power_law_1.01,2.3346431732177733
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,128,power_law_1.2,0.15801600217819214
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,128,power_law_1.01,0.03251200020313263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,128,power_law_1.2,0.17787519693374634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,128,power_law_1.2,0.16531200408935548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,128,power_law_1.2,0.18791040182113647
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,128,power_law_1.01,0.03473919928073883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,128,power_law_1.01,1.9768640518188476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,128,power_law_1.2,0.17172479629516602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,128,power_law_1.2,0.2092288017272949
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,128,power_law_1.01,0.046675199270248414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,128,power_law_1.2,0.1794368028640747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,128,power_law_1.2,0.23960320949554442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,128,power_law_1.2,0.1880959987640381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,128,power_law_1.01,0.07499520182609558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,128,power_law_1.2,0.27524480819702146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,128,power_law_1.2,0.2038655996322632
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,128,power_law_1.01,0.07427200078964233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,128,power_law_1.2,0.3523583889007568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,128,power_law_1.2,0.21891839504241944
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,128,power_law_1.01,0.07296000123023987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,128,power_law_1.2,0.24597759246826173
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,128,power_law_1.2,0.16717439889907837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,128,power_law_1.2,0.4215424060821533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,128,balanced,0.02749866743882497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,128,power_law_1.01,0.05702400207519531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,128,balanced,0.02717333287000656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,128,balanced,0.02743999908367793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,128,balanced,0.029194665451844532
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,128,power_law_1.2,0.17885440587997437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,128,balanced,0.029077333708604176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,128,power_law_1.01,0.04766719937324524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,128,power_law_1.2,0.5643392086029053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,128,power_law_1.2,0.18765439987182617
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,128,power_law_1.01,0.05079039931297302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,128,power_law_1.2,0.2105855941772461
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,128,power_law_1.2,0.7127552032470703
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,128,power_law_1.01,0.050393599271774295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,128,power_law_1.2,0.24053120613098145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,128,power_law_1.01,0.05145599842071533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,128,power_law_1.2,0.27934720516204836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,128,power_law_1.2,1.0071807861328126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,128,power_law_1.01,0.048390400409698484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,128,power_law_1.01,2.96060791015625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,128,power_law_1.2,0.35651841163635256
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,128,power_law_1.01,0.051648002862930295
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,128,power_law_1.2,0.43695359230041503
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,128,power_law_1.01,0.06382079720497132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,128,power_law_1.2,1.2981823921203612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,128,power_law_1.2,0.31388800144195556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,128,balanced,0.029391999046007793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,128,balanced,0.033514666060606636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,128,power_law_1.2,0.5824384212493896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,128,power_law_1.01,0.07359359860420227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,128,balanced,0.03356266766786575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,128,balanced,0.03743999948104223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,128,balanced,0.03731200098991394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,128,power_law_1.2,0.3895999908447266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,128,balanced,0.04342400034268697
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,128,power_law_1.01,0.09681919813156128
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,128,power_law_1.2,0.7293951988220215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,128,power_law_1.2,0.49729280471801757
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,128,power_law_1.01,0.11699839830398559
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,128,power_law_1.01,3.8740798950195314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,128,balanced,0.029232000311215717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,128,balanced,0.029146666328112285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,64,power_law_1.2,10.339046478271484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,128,power_law_1.2,1.0424063682556153
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,128,power_law_1.01,0.16087679862976073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,128,power_law_1.2,0.6273791790008545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,128,power_law_1.01,0.20634880065917968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,128,power_law_1.2,0.8807680130004882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,128,power_law_1.2,1.3401151657104493
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,128,power_law_1.01,0.30074880123138426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,128,power_law_1.2,0.02680320143699646
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,128,power_law_1.01,0.39733760356903075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,128,power_law_1.2,1.1448512077331543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,128,power_law_1.2,2.0617343902587892
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,128,power_law_1.01,0.5444672107696533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,128,power_law_1.2,1.9059904098510743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,128,power_law_1.2,1.6612224578857422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,128,power_law_1.01,0.727449607849121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,128,power_law_1.2,2.6976512908935546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,128,power_law_1.2,2.5189888000488283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,128,power_law_1.2,2.2075712203979494
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,128,power_law_1.2,0.02682879865169525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,128,power_law_1.01,0.02600319981575012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,128,power_law_1.01,0.026393601298332216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,128,power_law_1.01,0.0263808012008667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,128,power_law_1.01,0.026752001047134398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,128,power_law_1.01,0.02629759907722473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,128,power_law_1.01,0.02632960081100464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,128,power_law_1.01,0.026598399877548216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,128,power_law_1.01,0.026694399118423463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,128,balanced,0.02956799914439519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,128,balanced,0.029285334050655365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,128,balanced,0.02924266705910365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,128,balanced,0.025439999997615814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,128,power_law_1.01,0.027142399549484254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,128,balanced,0.03160000095764796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,128,balanced,0.023631999890009563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,128,balanced,0.03365866591533025
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,128,balanced,0.02365333338578542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,128,power_law_1.2,3.249043273925781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,128,balanced,0.035631999373435974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,128,power_law_1.01,0.029407998919486998
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,128,balanced,0.03773866593837738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,128,balanced,0.0377813329299291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,128,balanced,0.045893331368764244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,128,balanced,0.04717333118120829
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,128,balanced,0.04794666667779287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,128,balanced,0.048309331138928734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,128,balanced,0.05178666611512502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,128,balanced,0.052186667919158936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,128,balanced,0.05793066819508871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,128,balanced,0.06192000210285187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,128,balanced,0.06851733227570851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,128,balanced,0.07643199960390727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,128,balanced,0.08902399738629659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,128,balanced,0.11955199639002483
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,128,power_law_1.2,0.02794879972934723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,128,balanced,0.145306666692098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,128,balanced,0.20545067389806113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,128,balanced,0.2600266734759013
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,128,power_law_1.2,0.02956799864768982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,128,balanced,0.3635520140329997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,128,power_law_1.2,4.240934371948242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,128,balanced,0.46980265776316327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,128,power_law_1.01,0.027084800601005554
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,128,power_law_1.2,0.032364800572395325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,128,balanced,0.6779466470082601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,128,power_law_1.01,0.028787198662757873
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,128,balanced,0.025498665869235992
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,128,balanced,0.02756800005833308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,128,power_law_1.01,0.03212159872055054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,128,balanced,0.889135996500651
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,128,balanced,0.03364799916744232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,128,power_law_1.01,0.03150080144405365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,128,balanced,0.03364799916744232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,128,power_law_1.01,0.032627201080322264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,128,balanced,0.04906666775544485
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,128,balanced,0.03575466573238373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,128,power_law_1.01,0.03232640027999878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,128,balanced,0.05320000151793162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,128,balanced,0.035599999129772186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,128,power_law_1.01,0.03578880131244659
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,128,balanced,0.03570133447647095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,128,power_law_1.01,0.03656319975852966
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,128,balanced,0.03576533248027166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,128,power_law_1.01,0.03588480055332184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,128,balanced,0.03774933268626531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,128,power_law_1.01,0.03663359880447388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,128,power_law_1.01,0.04177919924259186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,128,power_law_1.01,0.04319359958171844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,128,power_law_1.01,0.04421760141849518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,128,power_law_1.01,0.047251200675964354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,128,power_law_1.01,0.04471679925918579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,128,power_law_1.01,0.048819199204444885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,128,power_law_1.01,0.0495743989944458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,128,power_law_1.01,0.05177599787712097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,128,power_law_1.01,0.051686400175094606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,128,power_law_1.01,0.055904000997543335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,128,power_law_1.01,0.03523840010166168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,128,power_law_1.01,0.056627202033996585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,128,power_law_1.2,0.03457919955253601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,128,power_law_1.01,0.0625216007232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,128,power_law_1.01,0.03558399975299835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,128,power_law_1.01,0.02418559938669205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,128,power_law_1.01,0.07152000069618225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,128,power_law_1.01,0.03534719944000244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,128,power_law_1.2,0.04705919921398163
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,128,power_law_1.01,0.023571200668811798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,128,balanced,0.056133334835370384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,128,power_law_1.01,0.08228480219841003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,128,balanced,0.06329066554705302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,128,power_law_1.01,0.03585920035839081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,128,balanced,0.07045333087444305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,128,power_law_1.01,0.02396800071001053
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,128,power_law_1.2,0.07416960000991821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,128,balanced,0.08204266428947449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,128,power_law_1.01,0.11029759645462037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,128,balanced,0.03770133356253306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,128,power_law_1.01,0.03642880022525787
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,128,balanced,0.03557866563399633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,128,power_law_1.01,0.025209599733352663
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,128,balanced,0.03766933331886927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,128,power_law_1.01,0.13651200532913207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,128,power_law_1.01,0.038227200508117676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,128,power_law_1.2,0.07300480008125305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,128,power_law_1.01,0.026918399333953857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,128,power_law_1.01,0.19132800102233888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,128,power_law_1.01,0.03857280015945434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,128,power_law_1.2,0.07240319848060608
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,128,power_law_1.01,0.03258239924907684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,128,power_law_1.01,0.039263999462127684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,128,power_law_1.01,0.2435904026031494
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,128,power_law_1.01,0.03375360071659088
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,128,power_law_1.01,0.049644801020622256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,128,power_law_1.01,0.34350080490112306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,128,power_law_1.01,0.034867200255393985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,128,power_law_1.01,0.06113920211791992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,128,power_law_1.01,0.44641919136047364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,128,power_law_1.01,0.03463680148124695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,128,power_law_1.01,0.06858879923820496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,128,power_law_1.01,0.0351936012506485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,128,power_law_1.01,0.07626240253448487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,128,power_law_1.01,0.6443264007568359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,128,power_law_1.01,0.03546879887580871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,128,power_law_1.01,0.09136000275611877
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,128,balanced,0.10828266541163127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,128,power_law_1.01,0.034860798716545106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,128,balanced,0.12995200355847678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,128,power_law_1.01,0.12343039512634277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,128,power_law_1.01,0.8552639961242676
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,128,balanced,0.03766400118668874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,128,balanced,0.1801813244819641
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,128,power_law_1.01,0.03561600148677826
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,128,balanced,0.03765333443880081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,128,balanced,0.22804800669352213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,128,power_law_1.01,0.14995839595794677
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,128,power_law_1.01,0.03550080060958862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,128,balanced,0.3190773328145345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,128,power_law_1.01,0.21345279216766358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,128,balanced,0.40917332967122394
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,128,power_law_1.01,0.0361407995223999
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,128,power_law_1.2,0.05862399935722351
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,128,balanced,0.5904266834259033
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,128,power_law_1.01,0.03661440014839172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,128,power_law_1.01,0.27358078956604004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,128,power_law_1.01,0.05047680139541626
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,128,power_law_1.2,0.049772799015045166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,128,power_law_1.01,0.03733119964599609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,128,balanced,0.7718186378479004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,128,power_law_1.01,0.05280640125274658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,128,power_law_1.01,0.3846015930175781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,128,power_law_1.2,0.052249598503112796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,128,power_law_1.01,0.040268799662590025
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,128,power_law_1.01,0.05413119792938233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,128,power_law_1.2,0.02587519884109497
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,128,power_law_1.01,0.04284160137176514
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,128,power_law_1.2,0.0494271993637085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,128,power_law_1.01,0.4956352233886719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,128,power_law_1.01,0.05605120062828064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,128,power_law_1.2,0.026105600595474242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,128,power_law_1.01,0.04779520034790039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,128,power_law_1.2,0.04896000027656555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,128,power_law_1.01,0.05971840023994446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,128,power_law_1.2,0.026169601082801818
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,128,power_law_1.01,0.05249919891357422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,128,power_law_1.01,0.0643775999546051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,128,power_law_1.2,0.026335999369621277
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,128,power_law_1.2,0.04836480021476745
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,128,balanced,0.041077333192030586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,128,power_law_1.01,0.07022719979286193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,128,power_law_1.2,0.02743679881095886
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,128,balanced,0.04197866717974345
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,128,power_law_1.2,0.05334399938583374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,128,balanced,0.047872001926104225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,128,power_law_1.2,0.0286655992269516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,128,power_law_1.01,0.07648000121116638
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,128,balanced,0.05223466455936432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,128,balanced,0.062319998939832054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,128,power_law_1.2,0.06293759942054748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,128,power_law_1.2,0.03266560137271881
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,128,power_law_1.01,0.09621120095252991
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,128,power_law_1.2,0.03192960023880005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,128,power_law_1.01,0.11626240015029907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,128,power_law_1.2,0.036006399989128114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,128,power_law_1.01,0.15779199600219726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,128,power_law_1.2,0.03711360096931458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,128,power_law_1.01,0.1985983967781067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,128,power_law_1.2,0.04392319917678833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,128,power_law_1.01,0.2861311912536621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,128,power_law_1.01,0.7278016090393067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,128,power_law_1.01,0.3694400072097778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,128,power_law_1.2,0.02629759907722473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,128,power_law_1.2,0.02654080092906952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,128,power_law_1.01,0.523308801651001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,128,power_law_1.01,0.9454655647277832
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,128,power_law_1.2,0.026073598861694337
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,128,balanced,0.073253333568573
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,128,balanced,0.09865066409111023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,128,power_law_1.2,0.026790401339530943
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,128,power_law_1.2,0.07728000283241272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,128,power_law_1.01,0.6932672023773193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,128,power_law_1.2,0.027136000990867614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,128,power_law_1.2,0.1005247950553894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,128,power_law_1.2,0.028672000765800475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,128,power_law_1.2,0.0324864000082016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,128,power_law_1.01,1.0247296333312987
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,128,power_law_1.2,0.12306560277938842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,128,power_law_1.2,0.03244799971580505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,128,power_law_1.2,0.047539201378822324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,128,power_law_1.2,0.035743999481201175
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,128,power_law_1.2,0.17181440591812133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,128,power_law_1.2,0.04917120039463043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,128,power_law_1.2,0.0363072007894516
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,128,power_law_1.01,0.06815999746322632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,128,power_law_1.01,1.3364864349365235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,128,power_law_1.2,0.05188480019569397
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,128,power_law_1.2,0.04179840087890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,128,power_law_1.2,0.2219007968902588
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,128,power_law_1.01,0.08124160170555114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,128,power_law_1.2,0.0562175989151001
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,128,power_law_1.2,0.04405759871006012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,128,power_law_1.2,0.0610368013381958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,128,power_law_1.2,0.0462336003780365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,128,power_law_1.01,0.1161728024482727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,128,power_law_1.2,0.06885120272636414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,128,power_law_1.2,0.04930559992790222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,128,balanced,0.12808000048001608
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,128,power_law_1.01,0.15274239778518678
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,128,balanced,0.18623467286427817
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,128,power_law_1.2,0.07788159847259521
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,128,power_law_1.2,0.052883201837539674
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,128,balanced,0.24508267641067505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,128,power_law_1.2,0.05865600109100342
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,128,power_law_1.2,0.09164800047874451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,128,balanced,0.033904001116752625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,128,balanced,0.035599999129772186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,128,balanced,0.0373333344856898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,128,power_law_1.2,0.06361600160598754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,128,balanced,0.03572800010442734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,128,power_law_1.2,0.12369920015335083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,128,balanced,0.039749334255854286
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,128,balanced,0.03977066775163015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,128,power_law_1.2,0.07419520020484924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,128,balanced,0.04582933088143667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,128,balanced,0.04797333478927612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,128,balanced,0.052095999320348106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,128,power_law_1.2,0.08490880131721497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,128,balanced,0.05402133365472158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,128,power_law_1.2,0.15434880256652833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,128,balanced,0.06929066777229309
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,128,power_law_1.2,0.11514879465103149
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,128,power_law_1.2,0.023654399812221526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,128,power_law_1.2,0.14087040424346925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,128,power_law_1.2,0.023839999735355378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,128,power_law_1.2,0.22031359672546386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,128,power_law_1.2,0.2051392078399658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,128,power_law_1.2,0.33112320899963377
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,128,power_law_1.2,0.024128000438213348
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,128,power_law_1.2,0.26029438972473146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,128,power_law_1.2,0.02534399926662445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,128,power_law_1.2,0.43136000633239746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,128,power_law_1.2,0.026655998826026917
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,128,power_law_1.01,0.21452798843383789
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,128,power_law_1.2,0.03272959887981415
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,128,power_law_1.01,0.2838592052459717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,128,balanced,0.033674667278925575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,128,power_law_1.2,0.03409920036792755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,128,balanced,0.03532266616821289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,128,balanced,0.03366933266321818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,128,balanced,0.03365333378314972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,128,balanced,0.03575466573238373
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,128,balanced,0.03741333385308584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,128,balanced,0.07800533374150594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,128,balanced,0.08344533046086629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,128,balanced,0.08593066533406575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,128,balanced,0.08969600001970927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,128,balanced,0.10106133421262105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,128,power_law_1.2,0.2785151958465576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,128,balanced,0.11542399724324544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,128,balanced,0.1325493355592092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,128,balanced,0.15131733814875284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,128,balanced,0.1925386587778727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,128,balanced,0.2281279961268107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,128,balanced,0.30988800525665283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,128,balanced,0.38898134231567383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,128,balanced,0.5495626529057821
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,128,power_law_1.2,0.6149119853973388
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,128,balanced,0.7111413478851318
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,128,power_law_1.2,0.03527680039405823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,128,balanced,1.0354080200195312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,128,balanced,0.04180799921353658
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,128,power_law_1.2,0.8347264289855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,128,balanced,0.04354666670163473
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,128,power_law_1.2,0.035359999537467955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,128,balanced,0.050026665131251015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,128,balanced,0.0476693312327067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,128,balanced,0.052149335543314614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,128,balanced,0.046069333950678505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,128,balanced,0.06851199766000111
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,128,balanced,1.3615946769714355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,128,balanced,0.07285866638024648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,128,balanced,0.07865599791208903
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,128,balanced,0.08389866352081299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,128,balanced,0.09144000212351482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,128,balanced,0.09706667065620422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,128,balanced,0.11144000291824341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,128,balanced,0.12916800379753113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,128,balanced,0.1504693329334259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,128,balanced,0.20122132698694864
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,128,balanced,0.24660267432530722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,128,balanced,0.3385386864344279
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,128,balanced,0.4286880095799764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,128,balanced,0.6059733231862386
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,128,power_law_1.2,0.3956928014755249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,128,balanced,0.7849439779917399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,128,balanced,0.04586666822433472
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,128,balanced,0.045738667249679565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,128,balanced,0.0479360024134318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,128,balanced,0.055567999680837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,128,balanced,0.05741333464781443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,128,balanced,0.055071999629338585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,128,balanced,0.0762613316377004
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,128,balanced,0.07712533573309581
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,128,balanced,0.07855999966462453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,128,balanced,0.08759466807047527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,128,balanced,0.08761599659919739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,128,balanced,0.10353599985440572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,128,balanced,0.11752532919247945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,128,balanced,0.1258026659488678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,128,balanced,0.1442346672217051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,128,balanced,0.17135467131932577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,128,balanced,0.2018773357073466
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,128,balanced,0.02571200082699458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,128,balanced,0.27216533819834393
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,128,balanced,0.026191999514897663
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,128,balanced,0.02565866708755493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,128,balanced,0.3372639815012614
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,128,balanced,0.027424000203609467
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,128,balanced,0.029466666281223297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,128,balanced,0.4739840030670166
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,128,power_law_1.2,0.03592959940433502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,128,balanced,0.0371573343873024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,128,balanced,1.1449226538340251
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,128,balanced,0.03770133356253306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,128,balanced,0.6059253215789795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,128,balanced,0.039733332892258964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,128,power_law_1.2,0.035980799794197084
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,128,balanced,0.045925334095954895
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,128,balanced,0.045941332976023354
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,128,power_law_1.2,0.035180801153182985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,128,balanced,0.03993066648642222
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,128,balanced,0.876960039138794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,128,balanced,1.503551959991455
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,128,balanced,0.040821333726247154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,128,balanced,0.040693332751592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,128,power_law_1.2,0.036006399989128114
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,128,balanced,0.04091199984153112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,128,power_law_1.01,0.034508800506591795
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,128,balanced,0.04355733096599579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,128,balanced,1.1492853164672852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,128,balanced,0.041834667325019836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,128,power_law_1.2,0.036083200573921205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,128,balanced,0.04514666895071665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,128,power_law_1.2,0.5099775791168213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,128,balanced,0.049973333875338234
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,128,power_law_1.2,0.036345601081848145
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,128,balanced,0.05369600156943003
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,128,balanced,0.06251200040181477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,128,power_law_1.2,0.03738879859447479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,128,balanced,1.6890400250752766
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,128,power_law_1.2,0.037836799025535585
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,128,power_law_1.2,0.0416128009557724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,128,power_law_1.2,0.04297600090503693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,128,balanced,2.228970686594645
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,128,power_law_1.2,0.049107199907302855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,128,power_law_1.01,0.031200000643730165
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,128,power_law_1.2,0.0559935986995697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,128,power_law_1.01,0.03157120048999786
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,128,power_law_1.01,0.03195520043373108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,128,power_law_1.01,0.034643200039863584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,128,power_law_1.01,0.03264000117778778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,128,power_law_1.01,0.034745600819587705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,128,power_law_1.01,0.033580800890922545
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,128,balanced,0.07251200079917908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,128,power_law_1.01,0.03454079926013946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,128,balanced,0.09549333651860555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,128,power_law_1.01,0.0354559987783432
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,128,balanced,0.12591466307640076
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,128,power_law_1.01,0.03767040073871612
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,128,balanced,0.18107199668884277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,128,power_law_1.01,0.039801600575447085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,128,balanced,0.2384480039278666
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,128,balanced,0.33801599343617755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,128,power_law_1.2,0.3731584072113037
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,128,balanced,0.4349546829859416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,128,power_law_1.01,0.04126720130443573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,128,power_law_1.01,0.04129279851913452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,128,power_law_1.01,0.0452672004699707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,128,power_law_1.01,0.04888319969177246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,128,power_law_1.01,0.04560000002384186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,128,power_law_1.01,0.0496832013130188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,128,power_law_1.01,0.05189120173454285
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,128,power_law_1.01,0.06630399823188782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,128,power_law_1.01,0.05284479856491089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,128,power_law_1.01,0.07247359752655029
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,128,power_law_1.01,0.06484479904174804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,128,power_law_1.01,0.07683200240135193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,128,power_law_1.01,0.07516160011291503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,128,power_law_1.01,0.08339840173721313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,128,power_law_1.2,0.07234560251235962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,128,power_law_1.01,0.09065600037574768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,128,power_law_1.01,0.09919360280036926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,128,power_law_1.01,0.11458560228347778
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,128,power_law_1.01,0.13224960565567018
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,128,power_law_1.01,0.15522559881210327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,128,power_law_1.01,0.21374080181121827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,128,power_law_1.01,0.07991679906845092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,128,power_law_1.01,0.26153600215911865
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,128,power_law_1.01,0.08540800213813782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,128,power_law_1.01,0.09276800155639649
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,128,power_law_1.01,0.36450560092926027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,128,power_law_1.01,0.10026880502700805
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,128,power_law_1.2,0.08999040126800537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,128,power_law_1.01,0.46144638061523435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,128,power_law_1.01,0.11477760076522828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,128,power_law_1.01,0.13367040157318116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,128,power_law_1.01,0.6730815887451171
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,128,power_law_1.01,0.15287040472030639
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,128,power_law_1.01,0.20444159507751464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,128,power_law_1.01,0.8550080299377442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,128,power_law_1.01,0.25854079723358153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,128,power_law_1.01,0.3534208059310913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,128,power_law_1.01,1.2708864212036133
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,128,power_law_1.2,0.12578560113906861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,128,power_law_1.01,1.6668735504150392
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,128,power_law_1.01,0.44556798934936526
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,128,power_law_1.2,0.4764224052429199
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,128,power_law_1.01,0.654534387588501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,128,power_law_1.01,0.8427200317382812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,128,power_law_1.01,1.2701760292053224
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,128,power_law_1.2,0.16707839965820312
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,128,power_law_1.01,1.6484607696533202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,128,power_law_1.2,0.03461759984493255
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,128,power_law_1.01,0.024326400458812715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,128,power_law_1.2,0.03452160060405731
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,128,power_law_1.01,0.024140800535678863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,128,power_law_1.2,0.034118399024009705
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,128,power_law_1.01,0.024108800292015075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,128,power_law_1.2,0.034796801209449765
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,128,power_law_1.01,0.025785601139068602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,128,power_law_1.2,0.037459200620651244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,128,power_law_1.2,0.03984639942646027
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,128,power_law_1.01,0.02768000066280365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,128,power_law_1.2,0.044819200038909913
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,128,power_law_1.01,0.035174399614334106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,128,power_law_1.2,0.04565759897232056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,128,power_law_1.01,0.03683840036392212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,128,power_law_1.2,0.05256320238113403
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,128,power_law_1.01,0.03782399892807007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,128,power_law_1.2,0.05259519815444946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,128,power_law_1.01,0.04344319999217987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,128,power_law_1.2,0.0644864022731781
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,128,power_law_1.01,0.04297600090503693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,128,power_law_1.2,0.07315199971199035
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,128,power_law_1.01,0.0380160003900528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,128,power_law_1.2,0.08281599879264831
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,128,power_law_1.2,0.08638079762458802
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,128,power_law_1.2,0.0934656023979187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,128,power_law_1.2,0.10176000595092774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,128,power_law_1.2,0.1180351972579956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,128,power_law_1.2,0.13612159490585327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,128,power_law_1.2,0.16179200410842895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,128,power_law_1.2,0.21694719791412354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,128,power_law_1.2,0.2752000093460083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,128,power_law_1.01,0.0383679986000061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,128,power_law_1.01,0.03909760117530823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,128,power_law_1.2,0.39112319946289065
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,128,power_law_1.01,0.038073599338531494
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,128,power_law_1.01,0.040550398826599124
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,128,power_law_1.01,0.04135040044784546
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,128,power_law_1.01,0.04318079948425293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,128,power_law_1.01,0.04834559857845307
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,128,power_law_1.01,0.05368959903717041
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,128,power_law_1.01,0.06728960275650024
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,128,power_law_1.01,0.08062719702720642
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,128,power_law_1.01,0.11904640197753906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,128,power_law_1.01,0.14803839921951295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,128,power_law_1.01,0.20570240020751954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,128,power_law_1.01,0.27603199481964114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,128,power_law_1.2,0.4973504066467285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,128,power_law_1.01,0.3993024110794067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,128,power_law_1.01,0.5205056190490722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,128,power_law_1.2,0.03118079900741577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,128,power_law_1.2,0.03153280019760132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,128,power_law_1.2,0.03177599906921387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,128,power_law_1.2,0.03232640027999878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,128,power_law_1.2,0.0332863986492157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,128,power_law_1.2,0.03518719971179962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,128,power_law_1.2,0.040863999724388124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,128,power_law_1.2,0.04126720130443573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,128,power_law_1.2,0.04865919947624207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,128,power_law_1.2,0.04999679923057556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,128,balanced,0.03769599894682566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,128,balanced,0.037674665451049805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,128,power_law_1.2,0.0661184012889862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,128,power_law_1.2,0.07292159795761108
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,128,power_law_1.2,0.07813760042190551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,128,power_law_1.2,0.08414720296859741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,128,power_law_1.2,0.09081599712371827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,128,power_law_1.2,0.09912319779396057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,128,power_law_1.2,0.11619839668273926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,128,power_law_1.2,0.1333567976951599
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,128,power_law_1.2,0.15884159803390502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,128,balanced,0.037445334096749626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,128,balanced,0.037402667105197906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,128,balanced,0.03812800099452337
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,128,balanced,0.039642666776975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,128,balanced,0.03955733279387156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,128,balanced,0.04147200038035711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,128,balanced,0.05148266752560934
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,128,power_law_1.2,0.21913599967956543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,128,power_law_1.2,0.7010496139526368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,128,power_law_1.2,0.2725888013839722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,128,power_law_1.2,0.3715775966644287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,128,power_law_1.2,0.48479361534118653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,128,balanced,0.05236800014972687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,128,balanced,0.05201066533724467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,128,balanced,0.05484800040721893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,128,balanced,0.05755733450253805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,128,balanced,0.05862933397293091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,128,balanced,0.06411199768384297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,128,balanced,0.06757333377997081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,128,balanced,0.07715733349323273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,128,balanced,0.09051733215649922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,128,balanced,0.11225066582361858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,128,balanced,0.15012799700101218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,128,balanced,0.1872426668802897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,128,balanced,0.2648533384005229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,128,balanced,0.33672531445821124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,128,balanced,0.48307732741038006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,128,balanced,0.6271786689758301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,128,power_law_1.2,0.6931583881378174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,128,balanced,0.9147786299387614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,128,balanced,1.20414932568868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,128,power_law_1.2,0.04561919867992401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,128,power_law_1.2,0.045542401075363156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,128,power_law_1.2,0.046086400747299194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,128,power_law_1.2,0.046387198567390445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,128,power_law_1.2,0.046463999152183535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,128,power_law_1.2,0.7436607837677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,128,power_law_1.2,0.0525439977645874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,128,power_law_1.2,0.05265920162200928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,128,power_law_1.2,0.0591808021068573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,128,power_law_1.2,0.07584000229835511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,128,power_law_1.2,0.07516160011291503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,128,power_law_1.2,0.07998719811439514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,128,power_law_1.2,0.08673279881477355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,128,power_law_1.2,0.09418879747390747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,128,power_law_1.2,0.10361599922180176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,128,power_law_1.2,0.11676160097122193
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,128,power_law_1.2,0.12748160362243652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,128,power_law_1.2,0.14750720262527467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,128,power_law_1.2,0.1872383952140808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,128,power_law_1.2,0.2411263942718506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,128,power_law_1.2,0.3425600051879883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,128,power_law_1.2,0.4311679840087891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,128,power_law_1.01,0.044352000951766966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,128,power_law_1.01,0.04496639966964722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,128,power_law_1.2,0.6467711925506592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,128,power_law_1.01,0.04519039988517761
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,128,power_law_1.01,0.045414400100708005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,128,power_law_1.01,0.046003198623657225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,128,power_law_1.01,0.05411199927330017
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,128,power_law_1.2,0.23435521125793457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,128,power_law_1.01,0.05307520031929016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,128,power_law_1.01,0.05380480289459229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,128,power_law_1.01,0.07822080254554749
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,128,power_law_1.01,0.07980800271034241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,128,power_law_1.01,0.08042240142822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,128,power_law_1.01,0.08647680282592773
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,128,power_law_1.01,0.09392639994621277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,128,power_law_1.01,0.1071679949760437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,128,power_law_1.01,0.11477760076522828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,128,power_law_1.01,0.12567039728164672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,128,power_law_1.2,0.7415808200836181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,128,power_law_1.2,0.03460479974746704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,128,power_law_1.01,0.14506880044937134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,128,power_law_1.2,0.03460479974746704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,128,power_law_1.01,0.17586560249328614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,128,power_law_1.2,0.034969601035118106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,128,power_law_1.01,0.21918079853057862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,128,power_law_1.2,0.035097599029541016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,128,power_law_1.01,0.3130687952041626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,128,power_law_1.2,0.03496319949626923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,128,power_law_1.2,0.037401598691940305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,128,power_law_1.2,0.03811840116977692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,128,power_law_1.2,0.9068736076354981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,128,power_law_1.2,0.03860479891300202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,128,power_law_1.2,0.049209600687026976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,128,power_law_1.2,0.050547200441360476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,128,power_law_1.2,0.05149440169334411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,128,power_law_1.2,0.7996607780456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,128,power_law_1.2,0.054079997539520266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,128,power_law_1.2,0.054604798555374146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,128,power_law_1.2,0.059494400024414064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,128,power_law_1.2,0.9650752067565918
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,128,power_law_1.01,0.3851072072982788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,128,power_law_1.2,0.06517120003700257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,128,power_law_1.2,0.07179520130157471
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,128,power_law_1.01,0.5630911827087403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,128,power_law_1.2,0.07902719974517822
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,128,power_law_1.2,0.09717119932174682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,128,power_law_1.01,0.7253759860992431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,128,power_law_1.2,0.1187391996383667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,128,power_law_1.2,0.16699520349502564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,128,power_law_1.01,1.0401344299316406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,128,power_law_1.2,0.2109760046005249
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,128,power_law_1.2,0.3081536054611206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,128,power_law_1.01,1.4178367614746095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,128,power_law_1.2,0.38995840549468996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,128,power_law_1.2,0.5609663963317871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,128,power_law_1.01,2.0496448516845702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,128,power_law_1.01,2.7082239151000977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,128,power_law_1.2,0.9112192153930664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,128,balanced,0.04970133304595947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,128,balanced,0.04972266654173533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,128,balanced,0.04773333172003428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,128,balanced,0.04996266464392344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,128,balanced,0.05586666862169901
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,128,balanced,0.08683199683825175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,128,balanced,0.10977066556612651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,128,balanced,0.11242666840553284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,128,balanced,0.11827199657758077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,128,balanced,0.12148800492286682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,128,balanced,0.21370667219161987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,128,balanced,0.2266826629638672
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,128,balanced,0.23177599906921387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,128,balanced,0.24289600054423013
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,128,balanced,0.2520266572634379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,128,balanced,0.26265599330266315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,128,balanced,0.28886399666468304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,128,balanced,0.32515732447306317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,128,balanced,0.3556106487909953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,128,balanced,0.4164479970932007
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,128,balanced,0.4773013194402059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,128,balanced,0.5989919900894165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,128,balanced,0.7220746676127116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,128,balanced,0.9653333028157552
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,128,balanced,1.2435146967569988
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,128,power_law_1.2,0.024697600305080412
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,128,balanced,1.7333812713623047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,128,power_law_1.2,0.02436479926109314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,128,power_law_1.2,0.024889600276947022
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,128,power_law_1.2,0.026636800169944762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,128,balanced,2.2519893646240234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,128,balanced,0.0415786678592364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,128,balanced,0.04151466737190882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,128,power_law_1.2,0.028095999360084535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,128,power_law_1.2,0.03540480136871338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,128,power_law_1.2,0.03692159950733185
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,128,power_law_1.2,0.30734078884124755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,128,balanced,0.05807999769846598
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,128,power_law_1.2,0.03834879994392395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,128,balanced,0.060234665870666504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,128,balanced,0.060080001751581825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,128,balanced,0.06137066582838694
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,128,power_law_1.2,0.042054399847984314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,128,balanced,0.06225599845250448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,128,balanced,0.09077866872151692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,128,balanced,0.10106666882832845
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,128,power_law_1.2,0.04246399998664856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,128,balanced,0.10249066352844238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,128,balanced,0.23348265886306763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,128,power_law_1.2,0.038764798641204835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,128,balanced,0.23797333240509033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,128,balanced,0.2423893411954244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,128,balanced,0.2506399949391683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,128,power_law_1.2,1.1948160171508788
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,128,balanced,0.25692800680796307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,128,balanced,0.039493332306543984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,128,balanced,0.04166933397452036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,128,power_law_1.2,1.412268829345703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,128,balanced,0.04390400151411692
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,128,balanced,0.0649599979321162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,128,balanced,0.09687999884287517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,128,balanced,0.0965333382288615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,128,balanced,0.09401599566141765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,128,balanced,0.09617599844932556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,128,balanced,0.21151467164357504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,128,balanced,0.22582934300104776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,128,balanced,0.2302186687787374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,128,balanced,0.2424373428026835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,128,balanced,0.2506399949391683
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,128,balanced,0.26083733638127643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,128,balanced,0.28833599885304767
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,128,balanced,0.321343998114268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,128,balanced,0.3547999858856201
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,128,power_law_1.2,0.039084801077842714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,128,balanced,0.42288001378377277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,128,balanced,0.4892959992090861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,128,balanced,0.26893866062164307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,128,balanced,0.28299200534820557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,128,balanced,0.6254666646321615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,128,balanced,0.29390933116277057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,128,balanced,0.7600906689961752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,128,balanced,0.32264532645543414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,128,balanced,0.3733813365300496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,128,balanced,1.0272160371144612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,128,balanced,0.42402132352193195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,128,balanced,0.5216159820556641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,128,balanced,1.315551996231079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,128,balanced,0.6226133505503336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,128,balanced,0.8209866682688395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,128,balanced,1.8530880610148113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,128,balanced,1.0487573146820068
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,128,power_law_1.2,0.03959679901599884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,128,power_law_1.2,0.03966720104217529
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,128,balanced,1.44650665918986
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,128,power_law_1.2,0.04131200015544891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,128,power_law_1.2,0.7494400024414063
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,128,power_law_1.2,0.04188799858093262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,128,power_law_1.2,0.044249600172042845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,128,balanced,1.8731147448221843
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,128,power_law_1.2,0.04966399967670441
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,128,power_law_1.2,0.057030397653579715
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,128,power_law_1.2,0.07233920097351074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,128,balanced,2.6941652297973633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,128,power_law_1.2,0.09082880020141601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,128,balanced,2.40940793355306
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,128,power_law_1.2,0.12442879676818848
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,128,balanced,3.547295888264974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,128,power_law_1.01,0.07055360078811646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,128,power_law_1.01,0.06969599723815918
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,128,power_law_1.2,0.16405760049819945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,128,power_law_1.01,0.07067520022392274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,128,power_law_1.01,0.07226240038871765
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,128,power_law_1.01,0.07703040242195129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,128,power_law_1.01,0.0933184027671814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,128,power_law_1.01,0.10847359895706177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,128,power_law_1.01,0.11004799604415894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,128,power_law_1.2,1.5851712226867676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,128,power_law_1.01,0.11720960140228272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,128,power_law_1.2,1.3255743980407715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,128,power_law_1.01,0.1189568042755127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,128,power_law_1.01,0.21144959926605225
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,128,power_law_1.01,0.22478079795837402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,128,power_law_1.01,0.23161599636077881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,128,balanced,0.027456000447273254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,128,power_law_1.01,0.23859200477600098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,128,balanced,0.028624000648657482
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,128,balanced,0.02951466788848241
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,128,balanced,0.029648000995318096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,128,balanced,0.03362133353948593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,128,power_law_1.01,0.25093119144439696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,128,balanced,0.03713600089152654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,128,balanced,0.0525493323802948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,128,balanced,0.06004266440868378
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,128,balanced,0.11040533582369487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,128,power_law_1.01,0.25947520732879636
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,128,balanced,0.10717333356539409
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,128,balanced,0.10897599657376607
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,128,balanced,0.09286933143933614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,128,power_law_1.01,0.28614399433135984
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,128,balanced,0.0804746647675832
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,128,balanced,0.06691200037797292
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,128,balanced,0.05955733358860016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,128,balanced,0.062463998794555664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,128,power_law_1.01,0.3219264030456543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,128,balanced,0.06606400012969971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,128,power_law_1.2,1.8599424362182617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,128,power_law_1.01,0.364300799369812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,128,power_law_1.01,0.05943040251731872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,128,power_law_1.01,0.4384448051452637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,128,power_law_1.01,0.05937280058860779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,128,power_law_1.01,0.5167232036590577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,128,power_law_1.01,0.05978879928588867
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,128,power_law_1.01,0.061337602138519284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,128,power_law_1.01,0.654585599899292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,128,power_law_1.01,0.06146559715270996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,128,power_law_1.01,0.8136896133422852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,128,balanced,0.07656000057856242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,128,balanced,0.08665066957473755
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,128,balanced,0.105295995871226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,128,balanced,0.1251040001710256
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,128,power_law_1.01,1.0961088180541991
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,128,balanced,0.16473066806793213
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,128,balanced,0.20773333311080933
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,128,balanced,0.2919253309567769
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,128,balanced,0.3845173517862956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,128,power_law_1.01,1.3911423683166504
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,128,balanced,0.5394613345464071
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,128,balanced,0.7083893616994222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,128,power_law_1.2,0.23892478942871093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,128,power_law_1.01,0.07787520289421082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,128,power_law_1.01,0.07301759719848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,128,power_law_1.01,2.013484764099121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,128,power_law_1.01,0.07828480005264282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,128,power_law_1.01,0.09175040125846863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,128,power_law_1.01,0.07893120050430298
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,128,power_law_1.01,0.09708799719810486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,128,power_law_1.01,0.0791104018688202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,128,power_law_1.01,0.09258880019187928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,128,power_law_1.01,2.6498367309570314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,128,power_law_1.01,0.0770367980003357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,128,power_law_1.01,0.09543039798736572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,128,power_law_1.01,0.09637119770050048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,128,power_law_1.01,0.2100287914276123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,128,power_law_1.01,0.10051840543746948
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,128,power_law_1.01,0.2244863986968994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,128,power_law_1.01,0.23049600124359132
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,128,power_law_1.01,0.2388159990310669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,128,power_law_1.01,0.24990079402923585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,128,power_law_1.01,0.25980799198150634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,128,power_law_1.01,0.2866624116897583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,128,power_law_1.01,0.3172800064086914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,128,power_law_1.01,0.09999359846115112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,128,power_law_1.01,0.3542655944824219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,128,power_law_1.01,0.22977919578552247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,128,power_law_1.01,0.4316671848297119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,128,power_law_1.01,0.23556480407714844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,128,power_law_1.01,0.508019208908081
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,128,power_law_1.01,0.24081919193267823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,128,power_law_1.01,0.6602176189422607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,128,power_law_1.01,0.24773120880126953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,128,power_law_1.01,0.8073408126831054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,128,power_law_1.01,0.25077760219573975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,128,power_law_1.01,0.26654720306396484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,128,power_law_1.01,0.031430399417877196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,128,power_law_1.01,1.091801643371582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,128,power_law_1.01,0.2769727945327759
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,128,power_law_1.01,0.030739200115203858
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,128,power_law_1.01,0.2908224105834961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,128,power_law_1.01,1.3954879760742187
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,128,power_law_1.01,0.03080959916114807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,128,power_law_1.01,0.3176000118255615
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,128,power_law_1.01,0.03166080117225647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,128,power_law_1.01,0.38874239921569825
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,128,power_law_1.01,0.035104000568389894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,128,power_law_1.01,0.45131521224975585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,128,power_law_1.01,2.003596878051758
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,128,power_law_1.01,0.03869439959526062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,128,power_law_1.01,0.5763199806213379
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,128,power_law_1.01,0.05000320076942444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,128,power_law_1.01,0.6904384136199951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,128,power_law_1.01,2.5805055618286135
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,128,power_law_1.01,0.05764480233192444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,128,power_law_1.01,0.9472319602966308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,128,power_law_1.2,0.07065600156784058
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,128,power_law_1.01,0.09472000002861022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,128,power_law_1.2,0.0705407977104187
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,128,power_law_1.2,1.0988672256469727
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,128,power_law_1.01,0.09778559803962708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,128,power_law_1.01,1.2025343894958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,128,power_law_1.2,0.07088000178337098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,128,power_law_1.2,0.07299839854240417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,128,power_law_1.01,0.09504640102386475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,128,power_law_1.2,0.07546240091323853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,128,power_law_1.01,0.089683198928833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,128,power_law_1.2,0.09230719804763794
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,128,power_law_1.01,0.08170880079269409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,128,power_law_1.2,0.1056831955909729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,128,power_law_1.2,0.11038719415664673
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,128,power_law_1.2,0.11794559955596924
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,128,power_law_1.2,2.3327423095703126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,128,power_law_1.2,0.12011519670486451
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,128,power_law_1.01,1.6828224182128906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,128,power_law_1.2,0.2093440055847168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,128,power_law_1.2,0.22499840259552
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,128,power_law_1.01,0.06424959897994995
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,128,power_law_1.01,2.1545024871826173
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,128,power_law_1.2,0.31094400882720946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,128,power_law_1.2,0.2302783966064453
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,128,power_law_1.01,0.058278399705886844
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,128,power_law_1.2,0.23978240489959718
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,128,power_law_1.01,0.06144639849662781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,128,power_law_1.2,0.24800000190734864
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,128,power_law_1.01,0.0654911994934082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,128,power_law_1.2,0.259935998916626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,128,power_law_1.01,3.153049659729004
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,128,power_law_1.01,0.07345280051231384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,128,power_law_1.2,0.2913856029510498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,128,power_law_1.2,1.7413183212280274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,128,power_law_1.01,0.0857151985168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,128,power_law_1.2,0.33649919033050535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,128,power_law_1.2,0.9774335861206055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,128,power_law_1.2,0.058899199962615965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,128,power_law_1.2,0.3739135980606079
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,128,power_law_1.2,0.05915520191192627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,128,power_law_1.2,0.07765759825706482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,128,power_law_1.2,0.45957121849060056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,128,power_law_1.2,0.07896320223808288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,128,power_law_1.2,0.060192000865936277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,128,power_law_1.2,0.5389567852020264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,128,power_law_1.2,0.07859200239181519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,128,power_law_1.2,0.060550397634506224
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,128,power_law_1.2,0.6910336017608643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,128,power_law_1.2,0.07902079820632935
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,128,power_law_1.2,0.0597760021686554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,128,power_law_1.2,0.07987200021743775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,128,power_law_1.2,0.06716799736022949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,128,power_law_1.2,0.8460543632507325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,128,power_law_1.01,0.10999679565429688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,128,power_law_1.2,0.09419519901275634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,128,power_law_1.2,0.09404159784317016
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,128,power_law_1.01,0.1325824022293091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,128,power_law_1.2,1.2071935653686523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,128,power_law_1.2,0.0956928014755249
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,128,power_law_1.01,0.17741440534591674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,128,power_law_1.2,0.09645439982414246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,128,power_law_1.01,4.1580863952636715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,128,power_law_1.2,0.09941759705543518
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,128,power_law_1.01,0.2244800090789795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,128,power_law_1.2,0.092467200756073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,128,power_law_1.2,1.5188480377197267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,128,power_law_1.2,0.23272318840026857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,128,power_law_1.01,0.3277055978775024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,128,power_law_1.2,0.09422079920768738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,128,power_law_1.2,0.2362623929977417
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,128,power_law_1.01,0.43001599311828614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,128,power_law_1.2,0.21012480258941652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,128,power_law_1.2,0.24170880317687987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,128,power_law_1.2,2.208140754699707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,128,power_law_1.2,0.22430078983306884
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,128,power_law_1.01,0.6025983810424804
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,128,power_law_1.2,0.2483839988708496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,128,power_law_1.2,0.22801918983459474
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,128,power_law_1.2,0.25603840351104734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,128,power_law_1.2,0.2391808032989502
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,128,power_law_1.01,0.8083200454711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,balanced,0.03347733368476232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,128,power_law_1.2,0.2658623933792114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,balanced,0.03332266708215078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,balanced,0.03359466542800268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,128,power_law_1.2,2.8993152618408202
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,128,power_law_1.2,0.2502592086791992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,balanced,0.03427733232577642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,balanced,0.035546667873859406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,balanced,0.04577599962552389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,128,power_law_1.2,0.2805311918258667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,balanced,0.04582933088143667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,balanced,0.04571199913819631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,128,power_law_1.2,0.2609472036361694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,balanced,0.0498933345079422
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,balanced,0.0499893327554067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,balanced,0.08710400263468425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,balanced,0.09474666913350423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,balanced,0.09710400303204854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,128,power_law_1.2,0.2914432048797607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,balanced,0.10131733616193135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,balanced,0.10930132865905762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,balanced,0.11338667074839275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,balanced,0.12593600153923035
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,128,power_law_1.2,0.3230655908584595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,balanced,0.14202133814493814
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,balanced,0.15734400351842245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,balanced,0.19322667519251505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,128,power_law_1.2,0.3633536100387573
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,balanced,0.23044800758361816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,balanced,0.2991200089454651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,128,power_law_1.2,0.4414527893066406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,balanced,0.031290667752424874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,balanced,0.03338133295377096
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,128,power_law_1.2,0.5167488098144531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,128,power_law_1.2,0.29472639560699465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,128,power_law_1.2,0.6801727771759033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,128,power_law_1.2,0.3321791887283325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,128,power_law_1.2,0.028364801406860353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,128,power_law_1.2,0.8253696441650391
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,128,power_law_1.2,0.028787198662757873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,128,power_law_1.2,0.3899199962615967
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,128,power_law_1.2,0.03076480031013489
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,128,power_law_1.2,0.4697279930114746
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,balanced,0.3657653331756592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,128,power_law_1.2,1.141964817047119
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,balanced,0.5066560109456381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,128,power_law_1.2,0.031891199946403506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,128,power_law_1.2,0.5992000102996826
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,balanced,0.03156800071398417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,balanced,0.0332640012105306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,balanced,0.03189333279927572
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,balanced,0.037871999045213066
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,128,power_law_1.2,0.032467201352119446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,balanced,0.04358399907747904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,balanced,0.04206933577855428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,balanced,0.04782933493455251
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,128,power_law_1.2,0.7461184024810791
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,128,power_law_1.2,1.4558079719543457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,balanced,0.048058668772379555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,balanced,0.09097066521644592
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,128,power_law_1.2,0.03686400055885315
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,balanced,0.09744000434875488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,balanced,0.09923199812571208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,balanced,0.1053706705570221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,balanced,0.11356799801190694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,128,power_law_1.2,0.9868288040161133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,balanced,0.11763200163841248
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,128,power_law_1.2,0.04707840085029602
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,balanced,0.13056533535321554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,balanced,0.1460586686929067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,balanced,0.1665173371632894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,balanced,0.20564266045888266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,128,power_law_1.2,0.05414400100708008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,balanced,0.25031999746958417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,128,power_law_1.2,1.2828160285949708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,balanced,0.3272533416748047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,balanced,0.40559999148050946
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,128,power_law_1.2,0.09770879745483399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,balanced,0.641594648361206
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,balanced,0.5612959861755371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,balanced,0.9155519803365072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,128,power_law_1.2,0.09695360064506531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,balanced,0.7136747042338053
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,128,power_law_1.2,1.767616081237793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,balanced,1.0246826807657878
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,balanced,1.1921119689941406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,balanced,0.044256001710891724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,balanced,0.043696001172065735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,balanced,0.04399466514587402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,balanced,0.04381866753101349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,balanced,1.33460267384847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,balanced,0.043925335009892784
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,balanced,0.05619733532269796
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,balanced,0.05336533486843109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,balanced,0.05309333403905233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,balanced,0.0958133339881897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,128,power_law_1.2,2.1317888259887696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,128,power_law_1.2,2.371449661254883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,balanced,0.0972106655438741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,balanced,0.10111467043558757
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,balanced,0.10316266616185506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,balanced,0.10732799768447876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,balanced,0.11574932932853699
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,balanced,0.12385066350301106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,balanced,0.13192533453305563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,balanced,0.14825600385665894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,balanced,0.1783626675605774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,balanced,0.20779200394948324
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,balanced,0.26556267340977985
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,128,power_law_1.2,0.091212797164917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,128,power_law_1.2,2.7678783416748045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,balanced,0.3291520078976949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,128,power_law_1.2,3.393267059326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,balanced,0.4520159959793091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,power_law_1.01,0.03155840039253235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,128,power_law_1.2,0.08502399921417236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,balanced,0.5679306586583456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,power_law_1.01,0.03185279965400696
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,128,power_law_1.2,0.07118719816207886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,balanced,0.805402676264445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,power_law_1.01,0.03187839984893799
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,128,power_law_1.2,0.06183680295944214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,balanced,1.042138655980428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,power_law_1.01,0.03261440098285675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,balanced,0.0271573339899381
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,128,power_law_1.2,0.055871999263763426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,power_law_1.01,0.03391999900341034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,balanced,1.5201759338378906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,128,power_law_1.2,0.057734400033950806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,power_law_1.01,0.04431999921798706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,128,power_law_1.2,4.6242622375488285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,128,power_law_1.2,0.06161919832229614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,power_law_1.01,0.04471679925918579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,power_law_1.01,0.02974080145359039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,balanced,1.9943572680155437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,128,power_law_1.2,0.07408000230789184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,power_law_1.01,0.044633600115776065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,power_law_1.01,0.04255360066890716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,power_law_1.01,0.029951998591423036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,power_law_1.01,0.05576320290565491
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,128,power_law_1.2,0.08721280097961426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,power_law_1.01,0.030745598673820495
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,power_law_1.01,0.05602560043334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,128,power_law_1.2,0.11199359893798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,power_law_1.01,0.03078399896621704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,power_law_1.01,0.08819199800491333
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,128,power_law_1.2,0.1364351987838745
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,balanced,0.02682666728893916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,power_law_1.01,0.031615999341011045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,power_law_1.01,0.09336320161819459
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,balanced,0.025418666501839954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,balanced,0.027189334233601887
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,balanced,0.027477333943049114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,power_law_1.01,0.036447998881340024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,power_law_1.01,0.09626240134239197
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,balanced,0.03145066648721695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,balanced,0.03957866628964742
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,balanced,0.040261333187421165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,power_law_1.01,0.0406143993139267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,power_law_1.01,0.10189440250396728
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,balanced,0.041738669077555336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,balanced,0.04155199974775314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,power_law_1.01,0.10951679944992065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,power_law_1.01,0.04108799993991852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,128,power_law_1.2,0.4504000186920166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,power_law_1.01,0.04602240025997162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,power_law_1.01,0.11412479877471923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,power_law_1.01,0.04674560129642487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,power_law_1.01,0.12810879945755005
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,power_law_1.01,0.04207360148429871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,power_law_1.01,0.09009919762611389
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,power_law_1.01,0.13980799913406372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,power_law_1.01,0.0426367998123169
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,power_law_1.01,0.0955456018447876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,power_law_1.01,0.16301440000534057
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,power_law_1.01,0.0429504007101059
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,128,power_law_1.2,0.1838912010192871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,power_law_1.01,0.20341761112213136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,power_law_1.01,0.04376319944858551
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,128,power_law_1.2,0.24184319972991944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,power_law_1.01,0.2434175968170166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,power_law_1.01,0.05056639909744263
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,balanced,0.051872000098228455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,power_law_1.01,0.32314879894256593
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,balanced,0.04046933352947235
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,128,power_law_1.2,0.3531264066696167
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,balanced,0.041850666205088295
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,balanced,0.0415040006240209
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,balanced,0.04159466673930486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,power_law_1.01,0.4058432102203369
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,balanced,0.04370133578777313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,balanced,0.043968002001444496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,128,power_law_1.2,0.4541632175445557
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,balanced,0.04608533283074697
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,balanced,0.0481333335240682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,power_law_1.01,0.5623680114746094
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,balanced,0.056032001972198486
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,balanced,0.06443733473618825
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,balanced,0.08264000217119853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,balanced,0.10097066561381023
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,power_law_1.01,0.7014080047607422
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,balanced,0.15271466970443726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,power_law_1.01,0.09876480102539062
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,balanced,0.1955733299255371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,power_law_1.01,0.10538239479064941
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,power_law_1.01,0.11221120357513428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,power_law_1.01,1.029088020324707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,power_law_1.01,0.11806080341339112
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,power_law_1.01,0.05190399885177612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,power_law_1.01,0.13105920553207398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,power_law_1.01,0.05475839972496033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,power_law_1.01,0.14841599464416505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,power_law_1.01,1.3096575736999512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,power_law_1.01,0.09438080191612244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,power_law_1.01,0.16836479902267457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,power_law_1.01,0.09550719857215881
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,128,power_law_1.2,0.6615295886993409
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,power_law_1.01,0.09879680275917054
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,balanced,0.2797600030899048
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,power_law_1.01,0.10230400562286376
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,balanced,0.36056001981099445
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,128,power_law_1.2,0.9246848106384278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,power_law_1.01,0.10572799444198608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,power_law_1.2,0.03237760066986084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,power_law_1.2,0.029478400945663452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,power_law_1.01,0.11421439647674561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,power_law_1.2,0.032051199674606325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,power_law_1.2,0.02996479868888855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,power_law_1.01,0.12361600399017333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,power_law_1.2,0.032102400064468385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,power_law_1.2,0.030003198981285097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,power_law_1.01,0.13066879510879517
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,power_law_1.2,0.03260799944400787
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,power_law_1.2,0.030854400992393494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,power_law_1.01,0.15020159482955933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,power_law_1.01,0.20979840755462648
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,power_law_1.2,0.034329599142074584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,power_law_1.2,0.03194240033626557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,power_law_1.01,0.1847488045692444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,power_law_1.01,0.2575551986694336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,power_law_1.2,0.0447488009929657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,power_law_1.2,0.036051198840141296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,power_law_1.01,0.21585280895233155
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,power_law_1.01,0.3424191951751709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,power_law_1.2,0.04457600116729736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,power_law_1.2,0.0404992014169693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,power_law_1.01,0.28239998817443845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,power_law_1.01,0.4263552188873291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,power_law_1.2,0.045158401131629944
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,power_law_1.2,0.04097279906272888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,power_law_1.01,0.3612799882888794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,power_law_1.2,0.04758400022983551
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,power_law_1.2,0.04583039879798889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,power_law_1.2,0.04850560128688812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,power_law_1.01,0.5015168190002441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,power_law_1.2,0.04635519981384277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,power_law_1.2,0.0867904007434845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,power_law_1.2,0.08986240029335021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,power_law_1.01,0.6382207870483398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,power_law_1.2,0.09229440093040467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,power_law_1.2,0.09582080245018006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,power_law_1.2,0.09479039907455444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,power_law_1.2,0.098854398727417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,power_law_1.2,0.10072959661483764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,power_law_1.01,0.8945343971252442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,power_law_1.2,0.10496000051498414
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,power_law_1.01,0.023763200640678404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,power_law_1.2,0.11075199842453003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,power_law_1.2,0.11242239475250244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,power_law_1.2,0.11562240123748779
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,power_law_1.01,0.023731200397014617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,power_law_1.2,0.11751680374145508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,power_law_1.01,1.1750207901000977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,power_law_1.2,0.12865920066833497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,power_law_1.2,0.13230719566345214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,power_law_1.01,0.5916672229766846
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,power_law_1.01,0.023571200668811798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,power_law_1.2,0.14372479915618896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,power_law_1.2,0.14847359657287598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,power_law_1.2,0.16352640390396117
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,power_law_1.01,0.024063999950885772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,power_law_1.01,0.7575424194335938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,power_law_1.2,0.1698815941810608
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,power_law_1.01,1.69683837890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,power_law_1.2,0.20899839401245118
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,power_law_1.01,0.02598400115966797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,power_law_1.2,0.21263999938964845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,power_law_1.2,0.25119359493255616
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,power_law_1.01,0.028595200181007384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,power_law_1.2,0.26312320232391356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,power_law_1.2,0.33651840686798096
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,power_law_1.01,0.037536001205444335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,power_law_1.2,0.34865920543670653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,power_law_1.01,2.21014404296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,power_law_1.2,0.42719359397888185
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,power_law_1.01,0.03820799887180328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,power_law_1.2,0.4332223892211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,power_law_1.2,0.04249599874019623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,power_law_1.2,0.5849728107452392
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,power_law_1.01,0.03849599957466125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,power_law_1.2,0.6040832042694092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,power_law_1.2,0.042208001017570496
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,power_law_1.01,0.03891839981079102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,power_law_1.2,0.74966402053833
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,power_law_1.2,0.04287360012531281
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,power_law_1.2,0.7907648086547852
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,power_law_1.01,0.04872959852218628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,power_law_1.2,0.04294399917125702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,power_law_1.2,1.1133503913879395
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,power_law_1.01,0.0381056010723114
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,power_law_1.01,1.0785344123840332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,power_law_1.2,0.04371840059757233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,power_law_1.2,1.136742401123047
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,power_law_1.01,0.03842560052871704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,power_law_1.2,0.05148800015449524
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,power_law_1.01,0.038438400626182555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,power_law_1.2,1.4373248100280762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,power_law_1.2,0.05194240212440491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,power_law_1.01,1.4138879776000977
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,power_law_1.01,0.03919999897480011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,power_law_1.2,1.490944004058838
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,power_law_1.2,0.05240319967269898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,power_law_1.01,0.03978239893913269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,power_law_1.2,0.09434880018234253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,power_law_1.2,0.09530879855155945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,power_law_1.2,0.0989184021949768
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,power_law_1.2,0.10263680219650269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,power_law_1.2,0.10599679946899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,power_law_1.2,0.11407999992370606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,power_law_1.2,0.1260159969329834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,power_law_1.2,0.132915198802948
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,power_law_1.01,0.041465601325035094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,power_law_1.2,0.1522752046585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,power_law_1.01,0.04490880072116852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,power_law_1.2,0.1896064043045044
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,power_law_1.2,0.025248000025749208
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,power_law_1.01,0.04797439873218536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,power_law_1.2,0.22344961166381835
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,power_law_1.2,0.025139200687408447
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,power_law_1.01,0.05813120007514953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,power_law_1.2,0.298419189453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,power_law_1.2,0.02462719976902008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,power_law_1.01,0.07046399712562561
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,power_law_1.2,0.025388801097869874
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,power_law_1.01,0.09378560185432434
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,power_law_1.2,0.02744320034980774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,power_law_1.01,0.1215872049331665
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,power_law_1.01,0.1680575966835022
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,power_law_1.2,0.02924799919128418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,power_law_1.01,0.22305281162261964
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,power_law_1.2,0.03781760036945343
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,power_law_1.2,0.03852800130844116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,power_law_1.01,0.3182336091995239
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,power_law_1.2,0.039443200826644896
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,power_law_1.01,0.4112448215484619
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,power_law_1.2,0.03944959938526153
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,power_law_1.2,0.37224318981170657
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,power_law_1.2,0.04801279902458191
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,power_law_1.2,0.5375552177429199
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,power_law_1.2,0.03903360068798065
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,power_law_1.2,0.03968639969825745
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,power_law_1.2,0.6605055809020997
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,power_law_1.2,0.03888640105724335
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,power_law_1.2,0.039961600303649904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,power_law_1.2,0.9386303901672364
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,power_law_1.2,0.04081920087337494
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,128,balanced,0.03137599925200144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,128,balanced,0.03345066557327906
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,power_law_1.2,0.04281600117683411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,128,balanced,0.031386665999889374
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,power_law_1.2,0.04609279930591583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,128,balanced,0.031856000423431396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,128,balanced,0.03195200115442276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,128,balanced,0.031285333136717476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,128,balanced,0.031658666829268135
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,power_law_1.2,0.05003520250320435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,128,balanced,0.03551466763019562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,128,balanced,0.037717332442601524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,128,balanced,0.039503999054431915
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,power_law_1.2,0.06062080264091492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,128,balanced,0.043653334180514015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,128,balanced,0.045893331368764244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,128,balanced,0.06189866860707601
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,power_law_1.2,0.07444480061531067
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,power_law_1.2,0.09813759922981262
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,power_law_1.2,0.12667520046234132
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,power_law_1.2,0.18753279447555543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,128,balanced,0.03322133421897888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,128,balanced,0.033386667569478355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,128,balanced,0.033344000577926636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,128,balanced,0.03552533437808355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,power_law_1.2,0.23320960998535156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,128,balanced,0.04139200101296107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,power_law_1.2,1.2626943588256836
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,power_law_1.2,0.34683520793914796
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,power_law_1.2,0.4547135829925537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,power_law_1.2,1.8915071487426758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,128,power_law_1.01,0.031891199946403506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,128,power_law_1.2,1.4560192108154297
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,128,power_law_1.01,0.031718400120735166
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,128,power_law_1.01,0.02959359884262085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,128,power_law_1.01,0.03209599852561951
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,128,power_law_1.01,0.03206399977207184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,128,power_law_1.01,0.030022400617599487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,128,power_law_1.01,0.032979199290275575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,128,power_law_1.01,0.030272001028060914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,power_law_1.2,2.550432014465332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,128,balanced,0.04136000076929728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,128,balanced,0.04667733112970988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,128,power_law_1.01,0.0350847989320755
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,128,balanced,0.04784533381462097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,128,balanced,0.056032001972198486
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,128,power_law_1.2,0.03218559920787811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,128,balanced,0.06204266846179962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,128,power_law_1.01,0.0406464010477066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,128,balanced,0.06410133341948192
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,128,balanced,0.06849066913127899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,128,power_law_1.2,0.03198080062866211
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,128,power_law_1.01,0.0406143993139267
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,128,balanced,0.07249066730340321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,128,balanced,0.07727466523647308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,128,power_law_1.01,0.04624640047550201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,128,power_law_1.2,0.03199360072612763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,128,power_law_1.01,0.04593920111656189
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,128,power_law_1.2,0.03261440098285675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,128,power_law_1.01,0.0561024010181427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,128,power_law_1.2,0.033030399680137636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,128,power_law_1.2,0.03470720052719116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,128,power_law_1.2,0.04053759872913361
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,128,balanced,0.06433066725730896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,128,power_law_1.2,0.04078719913959503
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,128,power_law_1.2,0.02937600016593933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,128,power_law_1.2,0.04627839922904968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,128,power_law_1.01,0.030239999294281006
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,128,power_law_1.2,0.03017599880695343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,128,power_law_1.2,0.046374401450157164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,128,power_law_1.01,0.03126400113105774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,128,power_law_1.2,0.029894399642944335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,128,power_law_1.2,0.055955201387405396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,128,power_law_1.01,0.03370879888534546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,128,power_law_1.2,0.030700799822807313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,128,balanced,0.08477866649627686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,128,balanced,0.09692800045013428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,128,power_law_1.2,0.031071999669075014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,128,balanced,0.11340266466140747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,128,balanced,0.14141333103179932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,128,power_law_1.2,0.033529600501060484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,128,balanced,0.16895467042922974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,128,balanced,0.23832533756891885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,128,power_law_1.2,0.037625598907470706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,128,balanced,0.2997013330459595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,128,power_law_1.2,0.03845120072364807
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,128,balanced,0.4177440007527669
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,128,power_law_1.2,0.044223999977111815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,128,balanced,0.06796266635258992
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,128,power_law_1.2,0.044710400700569156
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,128,balanced,0.07296533385912578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,128,balanced,0.07866666714350383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,128,power_law_1.2,0.05939199924468994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,128,balanced,0.0828000009059906
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,128,balanced,0.09096533060073853
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,128,balanced,0.10610666871070862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,128,balanced,0.12362666924794515
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,128,balanced,0.155130664507548
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,128,balanced,0.18920000394185385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,128,power_law_1.01,0.03806079924106598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,128,balanced,0.26494399706522626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,128,power_law_1.01,0.03858560025691986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,128,balanced,0.33644266923268634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,128,power_law_1.01,0.04428159892559051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,128,balanced,0.47037867705027264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,128,power_law_1.01,0.04475519955158234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,128,balanced,0.602997342745463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,128,power_law_1.01,0.05899519920349121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,128,balanced,0.5331626733144125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,128,balanced,0.8706080118815104
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,128,power_law_1.01,0.06348159909248352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,128,balanced,0.7707680066426595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,128,power_law_1.01,0.06643840074539184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,128,balanced,1.1387999852498372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,128,power_law_1.01,0.06058239936828613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,128,balanced,1.0061866442362468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,128,balanced,0.0433599998553594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,128,power_law_1.01,0.06232960224151611
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,128,balanced,0.04144000013669332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,128,balanced,0.04359999795754751
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,128,balanced,0.07038400073846181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,128,power_law_1.01,0.06913279891014099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,128,balanced,0.1035093367099762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,128,balanced,0.1297920048236847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,128,balanced,0.11614400148391724
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,128,balanced,0.1202186644077301
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,128,balanced,0.11814399560292561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,128,balanced,0.12180800239245097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,128,balanced,0.257477343082428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,128,balanced,0.27730133136113483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,128,balanced,0.2922293345133464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,128,balanced,0.31349867582321167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,128,balanced,0.3342026472091675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,128,power_law_1.2,0.06104320287704468
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,128,balanced,0.3591093222300212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,128,power_law_1.2,0.06294400095939637
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,128,balanced,0.4017599821090698
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,128,power_law_1.2,0.06952959895133973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,128,balanced,0.4654239813486735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,128,power_law_1.01,0.07238399982452393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,128,power_law_1.2,0.074508798122406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,128,balanced,0.5385919809341431
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,128,power_law_1.01,0.0771776020526886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,128,power_law_1.2,0.07904639840126038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,128,balanced,0.6789120038350424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,128,power_law_1.01,0.0821120023727417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,128,power_law_1.01,0.0738752007484436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,128,power_law_1.01,0.09228799939155578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,128,power_law_1.01,0.07827200293540955
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,128,power_law_1.01,0.10741759538650512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,128,power_law_1.01,0.08712319731712341
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,128,power_law_1.01,0.12478079795837402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,128,power_law_1.01,0.10030720233917237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,128,power_law_1.01,0.15994880199432374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,128,power_law_1.01,0.11727999448776245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,128,power_law_1.2,0.06394879817962647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,128,power_law_1.01,0.1955135941505432
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,128,power_law_1.01,0.1489856004714966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,128,power_law_1.2,0.06635519862174988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,128,power_law_1.01,0.1881152033805847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,128,power_law_1.01,0.2789504051208496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,128,power_law_1.2,0.07241600155830383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,128,power_law_1.01,0.26620800495147706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,128,power_law_1.2,0.07747200131416321
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,128,balanced,0.035760000348091125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,128,power_law_1.01,0.3560960054397583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,128,balanced,0.037733333806196846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,128,balanced,0.0397173340121905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,128,power_law_1.2,0.08259199857711792
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,128,power_law_1.01,0.332204794883728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,128,power_law_1.2,0.09048960208892823
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,128,balanced,0.8171093463897705
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,128,power_law_1.01,0.5013184070587158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,128,power_law_1.2,0.10444799661636353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,128,power_law_1.01,0.4762112140655518
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,128,balanced,1.1349013646443684
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,128,power_law_1.2,0.12163840532302857
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,128,power_law_1.01,0.6430912017822266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,128,power_law_1.01,0.6130112171173095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,128,power_law_1.2,0.1625983953475952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,128,balanced,1.4134880701700847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,128,power_law_1.2,0.20318078994750977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,128,power_law_1.01,0.9348544120788574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,128,power_law_1.01,0.8754688262939453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,128,power_law_1.2,0.2782912015914917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,128,balanced,1.9970293045043945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,128,power_law_1.2,0.367903995513916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,128,power_law_1.01,1.2028544425964356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,128,power_law_1.01,1.2270144462585448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,128,balanced,0.05633600056171417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,128,balanced,0.07043200234572093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,128,power_law_1.2,0.5288191795349121
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,128,balanced,0.08646399776140849
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,128,power_law_1.2,0.09322879910469055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,128,balanced,0.1002346674601237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,128,power_law_1.01,0.06997119784355163
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,128,balanced,0.10457066694895427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,128,balanced,2.5595626831054688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,128,balanced,0.09750399986902873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,128,power_law_1.2,0.10817919969558716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,128,balanced,0.10288000106811523
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,128,power_law_1.2,0.6960063934326172
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,128,balanced,0.259226659933726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,128,power_law_1.2,0.12698240280151368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,128,balanced,0.27906666199366253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,128,balanced,0.2943840026855469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,128,power_law_1.2,0.16590720415115356
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,128,balanced,0.31594133377075195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,128,balanced,0.34091198444366455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,128,power_law_1.2,0.20320000648498535
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,128,balanced,0.3691146771113078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,128,balanced,3.7824586232503257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,128,balanced,0.4203039805094401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,128,power_law_1.2,0.29161601066589354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,128,balanced,0.48342398802439374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,128,balanced,0.5629066626230875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,128,balanced,0.7226346333821615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,128,power_law_1.2,0.3785919904708862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,128,balanced,0.8801013628641764
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,128,power_law_1.2,0.5332863807678223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,128,power_law_1.01,0.06928640007972717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,128,balanced,4.941941261291504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,128,balanced,1.2131840387980144
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,128,power_law_1.2,0.6936511993408203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,128,power_law_1.01,0.07006719708442688
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,128,power_law_1.01,0.057625597715377806
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,128,balanced,1.5301225980122883
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,128,power_law_1.01,0.07180160284042358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,128,power_law_1.01,0.056435197591781616
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,128,balanced,0.052042668064435325
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,128,balanced,0.052426666021347046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,128,power_law_1.01,0.09466879963874816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,128,balanced,0.05412266651789347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,128,balanced,0.07366399963696797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,128,power_law_1.01,0.05766400098800659
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,128,balanced,0.08074133098125458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,128,balanced,0.09161067008972168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,128,balanced,0.09886399904886882
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,128,power_law_1.01,0.12663040161132813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,128,balanced,0.10362133383750916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,128,power_law_1.01,0.06004480123519897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,128,balanced,0.2790293296178182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,128,balanced,0.28625067075093585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,128,power_law_1.01,0.11251200437545776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,128,balanced,0.30128000179926556
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,128,power_law_1.01,0.0667136013507843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,128,balanced,0.3160746693611145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,128,power_law_1.01,0.11696640253067017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,128,balanced,0.3343840042750041
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,128,power_law_1.2,1.0220288276672362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,128,balanced,0.36524800459543866
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,128,power_law_1.01,0.08472319841384887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,128,balanced,0.39397335052490234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,128,power_law_1.01,0.11392639875411988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,128,balanced,0.4270346562067668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,128,power_law_1.01,0.09742079973220825
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,128,power_law_1.01,0.11930240392684936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,128,balanced,0.4945760170618693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,128,power_law_1.2,1.3211135864257812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,128,power_law_1.01,0.10055040121078491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,128,balanced,2.1725333531697593
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,128,balanced,0.6156800190607706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,128,power_law_1.01,0.25560319423675537
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,128,balanced,0.741983970006307
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,128,power_law_1.01,0.09664000272750854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,128,power_law_1.01,0.0711359977722168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,128,power_law_1.01,0.276147198677063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,128,power_law_1.01,0.10100480318069457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,128,balanced,1.0181972980499268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,128,power_law_1.01,0.2904831886291504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,128,power_law_1.01,0.25715200901031493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,128,balanced,1.2629493077596028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,128,power_law_1.01,0.07118719816207886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,128,power_law_1.01,0.3118144035339355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,128,power_law_1.01,0.27754878997802734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,128,power_law_1.01,0.07208960056304932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,128,power_law_1.01,0.3373248100280762
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,128,power_law_1.01,0.2922111988067627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,128,power_law_1.01,0.07442560195922851
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,128,power_law_1.01,0.36435201168060305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,128,power_law_1.01,0.4161087989807129
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,128,power_law_1.01,0.07568640112876893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,128,power_law_1.01,0.4841472148895264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,128,power_law_1.01,0.08883200287818908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,128,balanced,2.8019679387410483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,128,power_law_1.01,0.5654655933380127
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,128,power_law_1.01,0.09689599871635438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,128,power_law_1.01,0.10067839622497558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,128,power_law_1.01,0.7257343769073487
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,128,power_law_1.2,1.0240256309509277
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,128,power_law_1.01,0.3145792007446289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,128,power_law_1.01,0.27555840015411376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,128,power_law_1.01,0.8933440208435058
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,128,balanced,4.1346133550008135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,128,power_law_1.01,0.34023680686950686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,128,power_law_1.01,0.28488960266113283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,128,power_law_1.01,0.3719360113143921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,128,power_law_1.01,0.29534718990325926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,128,power_law_1.01,1.2125568389892578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,128,power_law_1.01,0.42827520370483396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,128,power_law_1.01,0.3141119956970215
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,128,power_law_1.01,0.49084157943725587
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,128,power_law_1.01,0.3324287891387939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,128,balanced,1.779802640279134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,128,power_law_1.01,1.5507583618164062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,128,balanced,5.4180482228597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,128,power_law_1.01,0.36338560581207274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,128,power_law_1.01,0.5787199974060059
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,128,power_law_1.2,1.388742446899414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,128,balanced,2.2701493899027505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,128,power_law_1.2,0.07323520183563233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,128,power_law_1.01,0.7482880115509033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,128,power_law_1.2,0.07342079877853394
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,128,power_law_1.01,2.204697608947754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,128,power_law_1.01,0.9163071632385253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,128,power_law_1.2,0.07359359860420227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,128,balanced,3.3052266438802085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,128,power_law_1.01,1.2538304328918457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,128,power_law_1.2,0.07436800003051758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,128,power_law_1.01,2.8628608703613283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,128,power_law_1.2,0.09077759981155395
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,128,power_law_1.01,0.4065663814544678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,128,power_law_1.01,1.6000703811645507
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,128,balanced,4.309664090474446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,128,power_law_1.2,0.12549760341644287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,128,power_law_1.01,0.4422016143798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,128,power_law_1.2,0.11528960466384888
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,128,power_law_1.2,0.11700479984283448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,128,power_law_1.01,0.5169023990631103
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,128,power_law_1.2,0.5791999816894531
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,128,power_law_1.2,0.11383039951324463
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,128,power_law_1.01,2.261759948730469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,128,power_law_1.01,0.6582079887390136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,128,power_law_1.01,4.212108612060547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,128,power_law_1.2,0.07167360186576843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,128,power_law_1.2,0.05846400260925293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,128,power_law_1.01,0.8011967658996582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,128,power_law_1.2,0.07170559763908387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,128,power_law_1.2,0.058259201049804685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,128,power_law_1.2,3.0791168212890625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,256,balanced,0.05006400247414907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,128,power_law_1.2,0.07343999743461609
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,128,power_law_1.01,2.9450815200805662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,128,power_law_1.01,1.0882944107055663
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,128,power_law_1.2,0.05912960171699524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,128,power_law_1.2,0.0729856014251709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,128,power_law_1.2,0.057651197910308837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,128,power_law_1.01,1.3665727615356444
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,128,power_law_1.2,0.07323520183563233
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,128,power_law_1.2,0.11978240013122558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,128,power_law_1.2,0.06469759941101075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,128,power_law_1.01,5.56682243347168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,128,power_law_1.2,0.25484800338745117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,128,power_law_1.2,0.09299200177192687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,256,balanced,0.04189866781234741
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,128,power_law_1.2,0.08272640109062195
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,256,balanced,0.041706666350364685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,256,balanced,0.03979733337958654
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,256,balanced,0.04192533095677694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,128,power_law_1.2,0.27472639083862305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,256,balanced,0.04474666714668274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,128,power_law_1.2,0.09633920192718506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,256,balanced,0.05389333268006643
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,128,power_law_1.2,0.09856640100479126
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,128,power_law_1.01,4.376947021484375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,256,balanced,0.06204266846179962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,256,balanced,0.05032533407211304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,256,balanced,0.06507200002670288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,128,power_law_1.2,0.2904256105422974
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,256,balanced,0.05013333261013031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,256,balanced,0.07057600220044453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,128,power_law_1.2,0.10172159671783447
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,128,power_law_1.2,0.10279680490493774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,256,balanced,0.07424533367156982
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,256,balanced,0.15060800313949585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,128,power_law_1.2,0.3185215950012207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,256,balanced,0.16426666577657065
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,128,power_law_1.2,0.0969983994960785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,256,balanced,0.1688800056775411
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,256,balanced,0.17971199750900269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,128,power_law_1.2,0.3416575908660889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,128,power_law_1.2,0.09699199795722961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,256,balanced,0.19313599665959677
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,256,balanced,0.20164267222086588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,128,power_law_1.2,0.25673599243164064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,128,power_law_1.2,0.37366399765014646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,128,balanced,6.391743977864583
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,128,power_law_1.2,0.27630081176757815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,128,power_law_1.01,1.9278079986572265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,128,power_law_1.2,0.4254591941833496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,128,power_law_1.2,0.29183359146118165
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,256,balanced,0.051354666550954185
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,128,power_law_1.2,0.49393281936645506
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,256,balanced,0.05589866638183594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,256,balanced,0.07226133346557617
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,128,power_law_1.01,5.665260696411133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,128,power_law_1.2,0.31763200759887694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,256,balanced,0.08135466774304707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,128,power_law_1.2,0.27652480602264407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,256,balanced,0.0843999981880188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,256,balanced,0.10524266958236694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,128,power_law_1.2,0.5839680194854736
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,256,balanced,0.10804800192515056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,256,balanced,0.16631999611854553
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,128,power_law_1.2,0.34277119636535647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,128,power_law_1.01,2.4931135177612305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,128,power_law_1.2,0.7575232028961182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,128,power_law_1.2,0.37781760692596433
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,256,balanced,0.2230559984842936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,256,balanced,0.25008533398310345
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,128,power_law_1.2,0.9179007530212402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,256,balanced,0.2856160004933675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,128,power_law_1.2,0.433292818069458
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,256,balanced,0.35443735122680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,128,balanced,8.414533615112305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,256,balanced,0.4219626585642497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,128,power_law_1.2,1.2514431953430176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,128,power_law_1.01,3.6403583526611327
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,128,power_law_1.2,0.2815615892410278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,128,power_law_1.2,1.6357311248779296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,128,power_law_1.2,0.2936959981918335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,128,power_law_1.2,0.3140928030014038
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,128,power_law_1.2,0.3334912061691284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,128,power_law_1.2,0.37683839797973634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,128,power_law_1.01,4.746976089477539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,256,balanced,0.5560640096664429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,128,power_law_1.2,0.4141632080078125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,256,balanced,0.692634661992391
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,256,balanced,0.965445359547933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,128,power_law_1.2,0.446284818649292
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,256,balanced,1.2363786697387695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,128,power_law_1.2,0.5327680110931396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,128,power_law_1.2,0.6636672019958496
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,256,balanced,1.7787893613179524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,128,power_law_1.2,2.290425682067871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,256,balanced,0.1749653418858846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,256,balanced,0.18132267395655313
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,256,balanced,0.1892426609992981
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,128,power_law_1.2,0.8257599830627441
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,256,balanced,0.19645333290100098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,256,balanced,2.3217172622680664
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,128,power_law_1.01,7.149900817871094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,128,power_law_1.2,1.1409472465515136
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,256,balanced,0.02924266705910365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,128,power_law_1.2,3.126016044616699
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,256,balanced,0.02951466788848241
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,256,balanced,0.029285334050655365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,256,balanced,0.031685332457224526
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,256,balanced,0.03334933271010717
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,128,power_law_1.2,1.4170495986938476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,128,power_law_1.2,0.49547519683837893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,256,balanced,0.20535467068354288
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,256,balanced,0.22241065899531046
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,256,balanced,0.2548533280690511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,256,power_law_1.01,0.049184000492095946
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,256,balanced,0.2809813419977824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,256,balanced,0.3384000062942505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,256,power_law_1.01,0.05043839812278748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,256,balanced,0.3981333176294963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,128,power_law_1.2,4.521728134155273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,256,balanced,0.5195093154907227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,256,power_law_1.01,0.04965760111808777
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,256,balanced,0.6436266501744589
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,256,balanced,0.04365866879622141
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,256,power_law_1.01,0.05043839812278748
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,256,balanced,0.05765333275000254
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,256,balanced,0.04576533536116282
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,256,balanced,0.8962026437123617
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,256,balanced,0.06015466650327047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,256,power_law_1.01,0.05498239994049072
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,256,balanced,0.05966933568318685
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,256,balanced,0.04596266647179922
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,256,balanced,0.04764799773693085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,256,power_law_1.01,0.07175040245056152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,256,balanced,1.13865065574646
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,128,power_law_1.2,0.5828351974487305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,256,power_law_1.01,0.07998080253601074
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,128,power_law_1.2,2.063916778564453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,256,balanced,1.6379520098368328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,256,power_law_1.01,0.08181120157241821
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,128,power_law_1.2,0.7544127941131592
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,256,power_law_1.01,0.1028223991394043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,256,power_law_1.01,0.105075204372406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,128,power_law_1.2,0.9302207946777343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,256,balanced,2.13810666402181
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,256,power_law_1.01,0.1592128038406372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,128,power_law_1.2,2.628153610229492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,256,power_law_1.01,0.040345600247383116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,256,power_law_1.01,0.0401856005191803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,128,power_law_1.01,9.346182250976563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,128,power_law_1.2,1.2684736251831055
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,256,balanced,0.04977599779764811
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,256,balanced,0.04900800188382467
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,256,power_law_1.01,0.04067200124263763
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,256,balanced,0.05013333261013031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,256,power_law_1.01,0.05928320288658142
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,256,balanced,0.05243733525276184
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,256,balanced,0.05667200187842051
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,256,power_law_1.01,0.04147840142250061
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,256,balanced,0.06304533282915752
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,256,balanced,0.07055466870466869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,128,power_law_1.2,1.6361919403076173
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,256,balanced,0.08905067046483357
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,256,power_law_1.01,0.04311679899692535
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,256,balanced,0.11089600125948589
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,128,power_law_1.2,6.233452987670899
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,256,balanced,0.15324800213178
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,128,power_law_1.2,3.9566463470458983
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,256,power_law_1.01,0.0536575973033905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,256,balanced,0.19638399283091226
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,256,balanced,0.28538666168848675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,256,power_law_1.01,0.0616320013999939
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,256,balanced,0.36764268080393475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,256,power_law_1.01,0.06567680239677429
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,128,power_law_1.2,2.3212863922119142
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,256,balanced,0.5267946720123291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,256,power_law_1.01,0.06821119785308838
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,256,balanced,0.687941312789917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,256,power_law_1.01,0.07264639735221863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,256,power_law_1.2,0.047993600368499756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,256,power_law_1.01,0.1482624053955078
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,256,power_law_1.01,0.06090239882469177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,256,power_law_1.2,0.04742400050163269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,256,power_law_1.01,0.06159999966621399
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,256,power_law_1.2,0.047603198885917665
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,128,power_law_1.2,3.0139904022216797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,256,power_law_1.01,0.06144639849662781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,256,power_law_1.2,0.048665601015090945
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,128,power_law_1.2,5.20294418334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,256,power_law_1.2,0.054745602607727054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,256,power_law_1.2,0.07116159796714783
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,256,power_law_1.2,0.0806656002998352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,256,power_law_1.2,0.08236799836158752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,256,power_law_1.2,0.1047935962677002
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,256,power_law_1.2,0.10725120306015015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,256,power_law_1.2,0.16363519430160522
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,256,power_law_1.2,0.17311359643936158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,256,power_law_1.2,0.17858560085296632
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,256,power_law_1.2,0.18639999628067017
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,256,power_law_1.2,0.19750399589538575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,256,power_law_1.01,0.025868800282478333
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,256,power_law_1.2,0.20521600246429444
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,256,power_law_1.01,0.02592639923095703
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,256,power_law_1.2,0.22413439750671388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,256,power_law_1.01,0.02590720057487488
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,256,power_law_1.2,0.2559679985046387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,256,power_law_1.2,0.29500160217285154
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,256,power_law_1.01,0.027987200021743774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,256,power_law_1.01,0.06259199976921082
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,256,power_law_1.2,0.3771712064743042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,256,power_law_1.01,0.030144000053405763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,256,power_law_1.01,0.07411839962005615
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,256,power_law_1.2,0.45840001106262207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,256,power_law_1.01,0.07493759989738465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,128,power_law_1.2,4.579199981689453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,256,power_law_1.01,0.07444480061531067
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,256,power_law_1.2,0.6200128078460694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,256,power_law_1.01,0.16622719764709473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,256,power_law_1.2,0.7957759857177734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,256,power_law_1.01,0.1665984034538269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,128,power_law_1.2,7.733612823486328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,256,power_law_1.01,0.17815040349960326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,256,power_law_1.2,1.1468416213989259
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,256,power_law_1.01,0.187116801738739
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,256,power_law_1.01,0.19409279823303222
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,256,power_law_1.01,0.04131200015544891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,256,power_law_1.2,1.4581567764282226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,256,power_law_1.01,0.20717439651489258
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,256,power_law_1.01,0.06178560256958008
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,256,power_law_1.01,0.21858561038970947
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,256,power_law_1.01,0.0434688001871109
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,256,power_law_1.01,0.23216640949249268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,256,power_law_1.2,2.2708608627319338
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,256,power_law_1.01,0.06330239772796631
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,256,power_law_1.01,0.25477759838104247
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,256,power_law_1.01,0.06282879710197449
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,256,power_law_1.01,0.3145983934402466
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,256,power_law_1.01,0.17408640384674073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,256,power_law_1.01,0.04394879937171936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,256,power_law_1.01,0.37303040027618406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,256,power_law_1.01,0.17854080200195313
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,256,power_law_1.01,0.04433920085430145
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,256,power_law_1.01,0.18677760362625123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,256,power_law_1.01,0.4940927982330322
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,256,power_law_1.2,3.033535957336426
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,256,power_law_1.01,0.047251200675964354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,256,power_law_1.01,0.19494400024414063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,256,power_law_1.2,0.03976320028305054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,256,power_law_1.01,0.20441598892211915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,256,power_law_1.01,0.6158080101013184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,128,power_law_1.2,10.314189147949218
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,256,power_law_1.01,0.221612811088562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,256,power_law_1.01,0.1615615963935852
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,256,power_law_1.01,0.8519680023193359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,256,power_law_1.01,0.25084800720214845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,256,power_law_1.2,0.0573311984539032
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,256,power_law_1.01,0.16768640279769897
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,256,power_law_1.01,0.28279039859771726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,256,power_law_1.2,0.05756160020828247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,256,power_law_1.01,0.17868160009384154
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,256,power_law_1.01,1.0859392166137696
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,256,power_law_1.01,0.35438721179962157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,256,power_law_1.2,0.05847679972648621
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,256,power_law_1.01,0.18942079544067383
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,256,power_law_1.01,0.42552962303161623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,256,power_law_1.2,0.058976000547409056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,256,power_law_1.01,0.2012415885925293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,256,power_law_1.01,0.04643200039863586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,256,power_law_1.2,0.059520000219345094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,256,power_law_1.01,1.5809727668762208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,256,power_law_1.01,0.5681344032287597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,256,power_law_1.2,0.04012799859046936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,256,power_law_1.2,0.07614719867706299
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,128,power_law_1.2,6.064992141723633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,256,power_law_1.01,0.7130815982818604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,256,power_law_1.2,0.04020479917526245
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,256,power_law_1.2,0.07548159956932068
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,256,power_law_1.2,0.04101119935512543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,256,power_law_1.01,2.040985679626465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,256,power_law_1.01,1.010035228729248
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,256,power_law_1.2,0.07655680179595947
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,256,power_law_1.2,0.042745599150657655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,256,power_law_1.2,0.15571199655532836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,256,power_law_1.2,0.05267840027809143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,256,power_law_1.2,0.15826560258865358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,256,power_law_1.01,1.309939193725586
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,256,power_law_1.2,0.06107519865036011
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,256,power_law_1.01,0.2221440076828003
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,256,power_law_1.2,0.16998399496078492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,256,power_law_1.2,0.0626688003540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,256,power_law_1.01,3.0249408721923827
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,256,power_law_1.01,0.04843519926071167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,256,power_law_1.2,0.18231040239334106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,256,power_law_1.01,0.2523519992828369
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,256,power_law_1.2,0.06810240149497986
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,256,power_law_1.01,1.8840255737304688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,256,power_law_1.01,0.04991999864578247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,256,power_law_1.2,0.19307520389556884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,256,power_law_1.01,0.2883519887924194
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,256,power_law_1.2,0.07023360133171082
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,256,power_law_1.01,0.05415679812431336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,256,power_law_1.2,0.148089599609375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,256,power_law_1.01,0.362611198425293
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,256,power_law_1.01,0.06312320232391358
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,256,power_law_1.2,0.1627776026725769
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,256,power_law_1.01,0.43870720863342283
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,256,power_law_1.01,0.069760000705719
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,256,power_law_1.2,0.16705280542373657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,256,power_law_1.01,2.458239936828613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,256,balanced,0.045850664377212524
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,256,power_law_1.01,4.109862518310547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,256,balanced,0.04597333570321401
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,256,power_law_1.01,0.0964352011680603
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,256,balanced,0.04565866788228353
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,256,balanced,0.03770666569471359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,256,balanced,0.04692799846331278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,256,balanced,0.03773866593837738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,256,balanced,0.051957334081331887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,256,balanced,0.03788266579310099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,256,balanced,0.06609599788983662
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,256,balanced,0.03772799919048945
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,256,power_law_1.01,0.12165759801864624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,256,balanced,0.04177066683769226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,256,balanced,0.048245335618654885
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,256,balanced,0.058042665322621666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,256,balanced,0.058037335673967995
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,256,power_law_1.01,0.16817280054092407
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,256,balanced,0.06287466486295064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,256,balanced,0.06601066887378693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,256,balanced,0.13825600345929465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,256,balanced,0.1502026617527008
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,256,power_law_1.01,0.22028799057006837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,256,power_law_1.2,0.20650880336761473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,256,balanced,0.1543786625067393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,256,balanced,0.16449066996574402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,256,balanced,0.17595199743906656
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,256,power_law_1.2,0.2224639892578125
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,256,power_law_1.01,0.31395840644836426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,256,balanced,0.1853653391202291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,256,balanced,0.20327999194463095
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,256,power_law_1.2,0.23323519229888917
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,256,balanced,0.22674665848414102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,256,balanced,0.2609120011329651
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,256,power_law_1.2,0.25793919563293455
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,256,power_law_1.01,0.5817024230957031
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,256,power_law_1.2,0.32999680042266843
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,256,balanced,0.07129066685835521
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,256,power_law_1.2,0.028044798970222475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,256,balanced,0.07221866647402446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,256,balanced,0.08654399712880452
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,256,power_law_1.01,0.7307968139648438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,256,balanced,0.08679466446240743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,256,power_law_1.2,0.39355518817901614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,256,balanced,0.13611732920010886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,256,balanced,0.15438933173815408
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,256,balanced,0.15847999850908914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,256,balanced,0.1666933298110962
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,256,balanced,0.1728853384653727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,256,power_law_1.01,1.0247296333312987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,256,balanced,0.18007999658584595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,256,balanced,0.19740800062815347
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,256,balanced,0.2223200003306071
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,256,balanced,0.2527359922726949
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,256,balanced,0.058186665177345276
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,256,balanced,0.060506666700045265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,256,balanced,0.30801600217819214
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,256,balanced,0.059978668888409935
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,256,power_law_1.01,0.4139200210571289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,256,balanced,0.061194668213526406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,256,balanced,0.06196266909440359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,256,power_law_1.01,1.3292287826538085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,256,balanced,0.3628053267796834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,256,power_law_1.2,0.17990399599075318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,256,balanced,0.0786293347676595
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,256,balanced,0.07632533212502797
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,256,balanced,0.47333868344624835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,256,balanced,0.07987200220425923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,256,power_law_1.2,0.19134080410003662
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,256,power_law_1.01,0.5984447956085205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,256,balanced,0.16061866283416748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,256,balanced,0.5845013459523519
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,256,balanced,0.17308266957600912
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,256,balanced,0.18309332927068075
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,256,balanced,0.814250628153483
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,256,balanced,0.1872640053431193
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,256,power_law_1.2,0.027820798754692077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,256,balanced,0.1978399952252706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,256,power_law_1.01,1.901740837097168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,256,balanced,0.20778665939966837
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,256,balanced,0.21947733561197916
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,256,balanced,1.0347786744435628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,256,power_law_1.2,0.5347839832305908
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,256,balanced,0.23116799195607504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,256,balanced,0.2541653315226237
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,256,balanced,0.30452799797058105
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,256,balanced,1.4864853223164876
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,256,balanced,0.3511199951171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,256,power_law_1.2,0.6602176189422607
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,256,balanced,0.4499146540959676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,256,balanced,0.3256053328514099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,256,power_law_1.01,2.5082048416137694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,256,balanced,0.5513919989267985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,256,balanced,1.9368319511413574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,256,balanced,0.38599467277526855
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,256,balanced,0.7574826876322428
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,256,balanced,0.9573173522949219
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,256,balanced,1.3633012771606445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,256,balanced,0.055914665261904396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,256,power_law_1.2,0.945958423614502
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,256,balanced,0.056159997979799904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,256,balanced,0.05750933289527893
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,256,balanced,0.05589333176612854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,256,balanced,1.7684319814046223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,256,balanced,0.509002685546875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,256,balanced,0.05624533196290334
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,256,power_law_1.2,0.201529598236084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,256,balanced,0.6334346532821655
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,256,power_law_1.2,0.22291839122772217
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,256,power_law_1.01,0.7954495906829834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,256,power_law_1.2,1.2466303825378418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,256,power_law_1.2,0.028172799944877626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,256,balanced,0.8810239632924398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,256,power_law_1.2,0.2534336090087891
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,256,power_law_1.01,0.04089600145816803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,256,balanced,2.5846452713012695
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,256,power_law_1.2,0.030144000053405763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,256,balanced,1.1263519922892253
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,256,power_law_1.2,0.2938751935958862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,256,power_law_1.01,0.040838399529457094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,256,power_law_1.2,1.7622400283813477
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,256,power_law_1.2,0.031788799166679385
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,256,power_law_1.01,0.04126079976558685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,256,power_law_1.2,0.37096319198608396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,256,balanced,1.6202880541483562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,256,power_law_1.01,0.04292480051517487
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,256,power_law_1.2,0.04169600009918213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,256,balanced,3.3987998962402344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,256,power_law_1.2,0.44586877822875975
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,256,power_law_1.01,0.04840959906578064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,256,power_law_1.2,0.055276799201965335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,256,power_law_1.01,0.036447998881340024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,256,balanced,0.06949333349863689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,256,power_law_1.01,0.06177279949188232
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,256,balanced,2.1161066691080728
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,256,balanced,0.07237866520881653
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,256,balanced,0.0735999991496404
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,256,power_law_1.2,2.356460762023926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,256,power_law_1.01,0.03591040074825287
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,256,balanced,0.14601600170135498
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,256,power_law_1.01,0.06806399822235107
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,256,balanced,0.1467626690864563
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,256,power_law_1.01,0.053472000360488894
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,256,balanced,0.027290667096773785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,256,balanced,0.15477333466211954
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,256,balanced,0.027258666853109997
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,256,power_law_1.01,0.036524799466133115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,256,balanced,0.16664000352223715
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,256,power_law_1.01,0.06948480010032654
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,256,balanced,0.027376001079877216
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,256,balanced,0.17258665959040323
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,256,balanced,0.029296000798543293
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,256,power_law_1.01,0.05330560207366943
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,256,balanced,0.18330667416254678
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,256,balanced,0.03145600110292435
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,256,power_law_1.01,0.037324801087379456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,256,power_law_1.01,0.08535040020942689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,256,balanced,0.19758933782577515
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,256,balanced,0.04154666761557261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,256,balanced,0.20677334070205688
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,256,balanced,0.04229333500067393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,256,power_law_1.01,0.054118400812149046
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,256,balanced,0.05832533538341522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,256,balanced,0.06005866825580597
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,256,power_law_1.01,0.05413119792938233
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,256,balanced,0.06003733476003011
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,256,balanced,0.044112001856168113
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,256,balanced,0.04433066646258036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,256,power_law_1.2,3.5010433197021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,256,balanced,0.047397335370381675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,256,balanced,0.047397335370381675
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,256,balanced,0.048384000857671104
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,256,balanced,0.04797333478927612
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,256,power_law_1.2,0.6003776073455811
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,256,balanced,0.05213333169619242
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,256,balanced,0.05799466868241628
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,256,power_law_1.2,0.04394879937171936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,256,balanced,0.06445333361625671
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,256,balanced,0.08053333560625713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,256,power_law_1.2,0.7626688003540039
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,256,balanced,0.09613333145777385
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,256,balanced,0.13804800311724344
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,256,balanced,0.17569599548975626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,256,power_law_1.2,1.08024320602417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,256,power_law_1.01,0.038822400569915774
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,256,balanced,0.25383466482162476
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,256,balanced,0.3284426728884379
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,256,power_law_1.01,0.04849919974803925
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,256,balanced,0.4665866692860921
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,256,power_law_1.01,0.05612159967422485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,256,power_law_1.01,0.055276799201965335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,256,power_law_1.2,1.3909184455871582
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,256,balanced,0.6096266508102417
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,256,power_law_1.01,0.07004160284996033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,256,power_law_1.01,0.06901760101318359
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,256,power_law_1.2,0.05603839755058289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,256,power_law_1.2,2.0485824584960937
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,256,power_law_1.01,0.07284479737281799
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,256,power_law_1.2,0.05569919943809509
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,256,power_law_1.01,0.14324480295181274
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,256,power_law_1.2,0.04450559914112091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,256,power_law_1.01,0.08442879915237426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,256,power_law_1.01,0.1444416046142578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,256,power_law_1.01,0.13557759523391724
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,256,power_law_1.2,0.04573439955711365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,256,power_law_1.01,0.15152640342712403
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,256,power_law_1.2,0.046675199270248414
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,256,power_law_1.2,2.707596778869629
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,256,power_law_1.01,0.1639423966407776
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,256,power_law_1.2,0.04824959933757782
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,256,power_law_1.01,0.05649920105934143
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,256,power_law_1.2,0.040550398826599124
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,256,power_law_1.01,0.17104640007019042
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,256,power_law_1.2,0.0494592010974884
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,256,power_law_1.01,0.06310399770736694
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,256,power_law_1.2,0.04077439904212952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,256,power_law_1.01,0.18303999900817872
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,256,power_law_1.2,0.05079039931297302
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,256,power_law_1.01,0.06409599781036376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,256,power_law_1.2,0.040678399801254275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,256,power_law_1.01,0.19452799558639527
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,256,power_law_1.2,0.05564799904823303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,256,power_law_1.01,0.1367743968963623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,256,power_law_1.2,0.04181120097637177
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,256,power_law_1.01,0.2068864107131958
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,256,power_law_1.2,0.0639743983745575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,256,power_law_1.2,0.047865599393844604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,256,power_law_1.01,0.2298367977142334
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,256,power_law_1.2,0.07478399872779846
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,256,power_law_1.2,0.060787200927734375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,256,power_law_1.01,0.15061119794845582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,256,power_law_1.01,0.28625919818878176
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,256,power_law_1.2,0.1063423991203308
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,256,power_law_1.2,0.06751360297203064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,256,power_law_1.01,0.15531519651412964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,256,power_law_1.01,0.34351999759674073
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,256,power_law_1.2,0.13062399625778198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,256,power_law_1.01,0.16387200355529785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,256,power_law_1.2,0.06873599886894226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,256,power_law_1.01,0.45068798065185545
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,256,power_law_1.01,0.1712767958641052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,256,power_law_1.2,0.0863103985786438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,256,power_law_1.01,0.18142720460891723
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,256,power_law_1.2,0.08471680283546448
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,256,power_law_1.01,0.5588031768798828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,256,power_law_1.01,0.19895679950714112
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,256,power_law_1.01,0.025331199169158936
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,256,power_law_1.2,0.1378432035446167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,256,balanced,0.2258239984512329
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,256,power_law_1.01,0.22741758823394775
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,256,balanced,0.2715733249982198
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,256,power_law_1.2,0.15070079565048217
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,256,power_law_1.01,0.7778751850128174
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,256,power_law_1.01,0.02515200078487396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,256,balanced,0.32173866033554077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,256,power_law_1.2,4.825382232666016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,256,power_law_1.01,0.2564608097076416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,256,power_law_1.2,0.15544320344924928
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,256,balanced,0.4164266586303711
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,256,power_law_1.01,0.025145599246025087
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,256,power_law_1.2,0.16307200193405152
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,256,balanced,0.5079946517944336
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,256,power_law_1.01,1.0097855567932128
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,256,power_law_1.01,0.027385601401329042
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,256,power_law_1.2,0.17319040298461913
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,256,balanced,0.6960373719533285
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,256,power_law_1.01,0.02887679934501648
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,256,power_law_1.2,0.1854591965675354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,256,balanced,0.8808533350626627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,256,power_law_1.01,1.4365119934082031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,256,power_law_1.01,0.039417600631713866
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,256,power_law_1.2,0.24186239242553711
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,256,balanced,1.2535839875539143
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,256,power_law_1.01,0.04082559943199158
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,256,power_law_1.2,0.34437758922576905
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,256,power_law_1.01,0.05602560043334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,256,balanced,1.6269653638203938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,256,power_law_1.01,1.8679872512817384
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,256,power_law_1.2,0.4670271873474121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,256,power_law_1.01,0.05729280114173889
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,256,power_law_1.01,0.3249216079711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,256,power_law_1.01,0.38682239055633544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,256,balanced,2.3758559226989746
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,256,power_law_1.2,0.6942080020904541
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,256,power_law_1.2,0.1815808057785034
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,256,power_law_1.01,0.5151616096496582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,256,power_law_1.2,0.20270719528198242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,256,power_law_1.01,2.781599998474121
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,256,power_law_1.2,0.9006784439086915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,256,power_law_1.2,0.23114240169525146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,256,power_law_1.01,0.6515071868896485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,256,power_law_1.01,0.14837759733200073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,256,balanced,3.125210762023926
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,256,power_law_1.2,0.26591360569000244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,256,power_law_1.2,0.05435519814491272
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,256,power_law_1.01,0.15231360197067262
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,256,power_law_1.01,0.9085247993469239
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,256,power_law_1.2,0.054502397775650024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,256,power_law_1.01,0.16328959465026854
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,256,power_law_1.2,0.03653759956359863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,256,power_law_1.2,0.05485439896583557
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,256,power_law_1.01,0.17395199537277223
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,256,power_law_1.01,1.1703359603881835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,256,power_law_1.2,0.03619199991226196
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,256,power_law_1.01,0.05615360140800476
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,256,power_law_1.2,0.055180799961090085
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,256,power_law_1.01,0.18504320383071898
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,256,power_law_1.01,3.667161560058594
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,balanced,0.033530667424201965
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,256,power_law_1.2,0.03694080114364624
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,balanced,0.03297066688537598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,256,power_law_1.01,0.20389120578765868
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,256,power_law_1.01,0.042656001448631284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,256,power_law_1.2,0.03771519958972931
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,256,power_law_1.01,0.22883200645446777
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,256,power_law_1.01,0.042719998955726625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,256,power_law_1.01,1.6848960876464845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,256,power_law_1.2,0.03932160139083862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,256,power_law_1.01,0.26464641094207764
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,256,power_law_1.01,0.04475519955158234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,256,power_law_1.01,0.33297278881073
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,256,power_law_1.2,0.04848000109195709
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,256,power_law_1.01,0.04431999921798706
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,256,power_law_1.2,0.3410943984985352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,256,power_law_1.01,0.4010496139526367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,256,power_law_1.2,0.05546240210533142
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,256,power_law_1.01,0.04647040069103241
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,256,power_law_1.01,2.24719352722168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,256,power_law_1.2,0.417299222946167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,balanced,0.03134933362404505
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,256,power_law_1.2,0.056831997632980344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,256,power_law_1.01,0.5323967933654785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,256,power_law_1.2,0.559769582748413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,256,power_law_1.2,0.06305919885635376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,256,power_law_1.2,0.055731201171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,256,power_law_1.01,0.6679935932159424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,256,power_law_1.2,0.06394879817962647
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,256,power_law_1.2,0.7046144008636475
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,256,power_law_1.2,0.0693120002746582
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,256,power_law_1.2,0.13694080114364623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,256,power_law_1.01,0.9344511985778808
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,256,power_law_1.2,0.07215999960899352
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,256,power_law_1.2,1.0128704071044923
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,256,power_law_1.2,0.14833279848098754
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,256,power_law_1.2,0.07374079823493958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,256,power_law_1.2,0.15304960012435914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,256,power_law_1.01,1.2054335594177246
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,256,power_law_1.2,0.16302720308303834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,256,power_law_1.2,0.14246400594711303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,256,power_law_1.2,1.2940671920776368
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,256,power_law_1.2,0.17578879594802857
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,256,power_law_1.01,0.046777600049972536
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,balanced,0.031210665901501972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,256,power_law_1.2,0.14613759517669678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,balanced,0.03133333226044973
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,balanced,0.031530665854612984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,balanced,0.03143999973932902
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,256,power_law_1.2,0.027564799785614012
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,balanced,0.03751466671625773
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,256,power_law_1.01,0.0505407989025116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,balanced,0.03976533313592275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,balanced,0.033728001018365227
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,256,power_law_1.01,1.732659149169922
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,balanced,0.04152533411979675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,balanced,0.033514666060606636
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,balanced,0.033546666304270424
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,256,power_law_1.2,0.027136000990867614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,balanced,0.04363733530044556
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,256,power_law_1.01,0.05749760270118713
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,balanced,0.043791999419530235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,256,power_law_1.2,1.9751935958862306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,balanced,0.043696001172065735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,balanced,0.04794133206208547
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,256,power_law_1.2,0.027161601185798644
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,256,power_law_1.01,0.06380159854888916
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,256,power_law_1.01,0.0847487986087799
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,256,power_law_1.2,0.02895359992980957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,256,power_law_1.01,2.284614372253418
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,256,power_law_1.01,0.10899840593338013
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,256,power_law_1.2,0.03054719865322113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,balanced,0.04188266893227895
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,balanced,0.04187199970086416
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,balanced,0.04181333382924398
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,256,power_law_1.01,0.1523967981338501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,balanced,0.04376000165939331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,balanced,0.043765331308046974
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,256,power_law_1.2,0.040915200114250184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,balanced,0.04888000090916952
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,balanced,0.050106664498647056
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,256,power_law_1.01,0.19263360500335694
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,256,power_law_1.2,0.04176000058650971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,256,power_law_1.2,0.18479360342025758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,256,power_law_1.2,0.15353599786758423
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,256,power_law_1.01,0.28194561004638674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,balanced,0.04589866598447164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,256,power_law_1.2,0.2048896074295044
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,256,power_law_1.2,0.16320639848709106
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,256,power_law_1.2,0.05602560043334961
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,256,power_law_1.01,0.36627840995788574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,256,power_law_1.2,0.171014404296875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,balanced,0.04786133269468943
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,256,power_law_1.2,0.05508480072021484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,balanced,0.08681600292523702
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,balanced,0.09488000472386678
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,256,power_law_1.2,0.18437119722366332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,balanced,0.09562666217486064
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,256,power_law_1.2,0.055353599786758426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,balanced,0.10123733679453532
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,256,power_law_1.01,0.5305920124053956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,balanced,0.10729599992434184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,256,power_law_1.2,0.19661439657211305
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,balanced,0.11187199751536052
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,256,power_law_1.2,0.04325119853019714
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,balanced,0.12158399820327759
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,256,power_law_1.2,0.20755839347839355
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,256,power_law_1.2,2.6934335708618162
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,256,power_law_1.2,0.044064000248909
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,256,power_law_1.01,0.6902463912963868
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,256,power_law_1.2,0.23617920875549317
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,256,power_law_1.2,0.0444927990436554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,balanced,0.05031999945640564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,power_law_1.01,0.030604800581932066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,256,power_law_1.2,0.2952064037322998
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,256,power_law_1.2,0.045612800121307376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,power_law_1.01,0.03146879971027374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,balanced,0.045893331368764244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,256,power_law_1.2,0.04694400131702423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,256,power_law_1.2,0.35865600109100343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,power_law_1.01,0.031219199299812317
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,256,power_law_1.2,0.2286463975906372
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,256,power_law_1.2,0.04824320077896118
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,256,power_law_1.2,0.48439040184021
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,power_law_1.01,0.031711998581886294
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,256,power_law_1.2,0.05141760110855102
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,256,power_law_1.2,0.26873600482940674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,power_law_1.01,0.032678401470184325
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,256,power_law_1.2,0.05883520245552063
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,256,power_law_1.2,0.34161279201507566
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,power_law_1.01,0.042342400550842284
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,balanced,0.1362986663977305
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,256,power_law_1.2,0.06778879761695862
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,balanced,0.15371732910474142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,balanced,0.18690133094787598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,power_law_1.01,0.04286080002784729
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,256,power_law_1.2,0.4142911911010742
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,balanced,0.22801067431767783
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,256,power_law_1.2,0.09221119880676269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,balanced,0.09444266557693481
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,power_law_1.01,0.04337919950485229
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,balanced,0.09472533067067464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,balanced,0.09884799520174663
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,256,power_law_1.2,0.11757440567016601
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,256,power_law_1.2,0.5545087814331054
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,power_law_1.01,0.04684799909591675
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,balanced,0.09086933732032776
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,balanced,0.027280000348885853
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,256,power_law_1.2,0.1630336046218872
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,power_law_1.01,0.04705280065536499
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,256,power_law_1.2,0.6983424186706543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,power_law_1.01,0.08528000116348267
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,256,power_law_1.2,0.21047039031982423
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,power_law_1.01,0.09223679900169372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,256,power_law_1.2,0.608019208908081
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,256,power_law_1.2,0.31040000915527344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,256,power_law_1.2,0.9744576454162598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,power_law_1.01,0.09499520063400269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,power_law_1.01,0.10043519735336304
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,256,power_law_1.2,0.4079808235168457
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,256,power_law_1.2,0.8573311805725098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,power_law_1.01,0.10612479448318482
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,256,power_law_1.2,1.2735936164855957
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,balanced,0.2967626651128133
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,power_law_1.01,0.11048959493637085
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,256,power_law_1.2,0.6130047798156738
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,balanced,0.101200004418691
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,balanced,0.3627946774164836
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,balanced,0.10517332951227824
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,balanced,0.11313066879908244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,256,power_law_1.2,1.1129535675048827
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,balanced,0.4996906518936157
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,power_law_1.01,0.1219648003578186
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,balanced,0.1216319998105367
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,balanced,0.09821866949399312
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,balanced,0.027210667729377747
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,balanced,0.12771733601888022
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,balanced,0.1009173293908437
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,balanced,0.027290667096773785
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,power_law_1.01,0.1363711953163147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,balanced,0.14340800046920776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,balanced,0.10692266623179118
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,balanced,0.02812800059715907
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,balanced,0.6338506539662679
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,256,power_law_1.2,0.7811583995819091
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,balanced,0.11107200384140015
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,balanced,0.02922133356332779
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,balanced,0.17602666219075522
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,balanced,0.03604800005753835
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,balanced,0.11729600032170613
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,balanced,0.2035413384437561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,power_law_1.01,0.15448319911956787
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,balanced,0.037418665985266365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,balanced,0.12795199950536093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,balanced,0.9042452971140543
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,balanced,0.2649120092391968
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,balanced,0.03942933430274328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,balanced,0.14642133315404257
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,256,power_law_1.2,1.6151744842529296
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,power_law_1.01,0.1929792046546936
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,balanced,0.03945599993069967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,balanced,0.32681065797805786
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,balanced,0.039520000418027244
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,balanced,0.04384533564249674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,balanced,1.1767840385437012
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,balanced,0.0395413339138031
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,balanced,0.039781334499518074
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,balanced,0.03966933240493139
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,power_law_1.01,0.041331198811531064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,256,power_law_1.2,2.083980751037598
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,power_law_1.01,0.04134399890899658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,power_law_1.01,0.041388800740242
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,256,power_law_1.2,1.8672256469726562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,power_law_1.01,0.04141440093517303
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,power_law_1.01,0.03121280074119568
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,power_law_1.01,0.042105600237846375
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,balanced,0.16367999712626138
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,power_law_1.01,0.03078399896621704
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,256,power_law_1.2,3.1558847427368164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,power_law_1.01,0.04878079891204834
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,balanced,0.20318933327992758
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,power_law_1.01,0.2353663921356201
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,balanced,0.44604798158009845
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,balanced,0.2459519902865092
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,power_law_1.01,0.030950400233268737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,power_law_1.01,0.04947839975357056
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,balanced,0.562554677327474
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,balanced,0.04057066639264425
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,power_law_1.01,0.3105087995529175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,256,power_law_1.2,2.4838272094726563
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,balanced,0.04167466859022776
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,power_law_1.01,0.031001600623130798
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,power_law_1.01,0.0483711987733841
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,balanced,0.043605332573254905
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,balanced,0.798138697942098
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,balanced,0.045994664231936135
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,power_law_1.01,0.3862720012664795
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,power_law_1.01,0.032390400767326355
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,balanced,0.048698668678601585
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,power_law_1.01,0.09202560186386108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,balanced,0.055957332253456116
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,balanced,0.06243200103441874
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,balanced,1.0306453704833984
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,power_law_1.01,0.03663359880447388
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,balanced,0.07979733248551686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,power_law_1.01,0.5418816089630127
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,balanced,0.09686400492986043
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,balanced,0.14643200238545737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,power_law_1.01,0.04019840061664581
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,balanced,0.1912320057551066
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,balanced,1.5009973843892415
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,256,power_law_1.2,4.208108901977539
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,power_law_1.01,0.6884543895721436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,power_law_1.01,0.04023680090904236
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,power_law_1.2,0.03238399922847748
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,power_law_1.01,0.045747199654579164
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,balanced,1.969744046529134
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,power_law_1.01,0.9876543998718261
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,power_law_1.01,0.046387198567390445
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,balanced,0.3264586726824443
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,power_law_1.2,0.029785600304603577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,balanced,0.4041653474171956
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,power_law_1.01,0.08885759711265565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,power_law_1.2,0.029894399642944335
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,balanced,0.555840015411377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,power_law_1.01,0.09605119824409485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,power_law_1.01,1.2943552017211915
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,power_law_1.2,0.03023360073566437
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,balanced,0.7087732950846354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,power_law_1.01,0.09889280200004577
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,power_law_1.2,0.04035840034484863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,power_law_1.2,0.030374398827552794
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,power_law_1.01,0.10503679513931274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,balanced,1.0164053440093994
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,power_law_1.2,0.04120959937572479
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,power_law_1.2,0.03188480138778686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,power_law_1.01,0.11066880226135253
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,balanced,0.2733440001805623
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,power_law_1.01,0.02444159984588623
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,power_law_1.2,0.0403903990983963
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,balanced,1.3226079940795898
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,balanced,0.35452266534169513
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,256,balanced,0.04134399940570196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,256,balanced,0.03978666663169861
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,power_law_1.01,0.024294400215148927
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,power_law_1.2,0.0418368011713028
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,power_law_1.2,0.031455999612808226
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,power_law_1.2,0.04245119988918304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,power_law_1.2,0.03245440125465393
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,power_law_1.2,0.04961279928684235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,power_law_1.2,0.03218559920787811
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,power_law_1.2,0.04764159917831421
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,power_law_1.01,0.09336320161819459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,power_law_1.2,0.05006080269813538
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,power_law_1.01,0.09663360118865967
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,power_law_1.2,0.09188479781150818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,power_law_1.01,0.10026880502700805
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,power_law_1.2,0.03594880104064942
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,power_law_1.2,0.09300479888916016
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,power_law_1.01,0.10343040227890014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,256,balanced,0.041482667128245033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,power_law_1.2,0.03978239893913269
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,256,balanced,0.06018133461475372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,256,balanced,0.08245866497357686
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,power_law_1.01,0.11129599809646606
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,256,balanced,0.10729599992434184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,256,balanced,0.09888533751169841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,power_law_1.2,0.039852800965309146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,256,balanced,0.10117333134015401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,power_law_1.01,0.11915520429611207
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,256,balanced,0.10384533802668254
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,256,balanced,0.10550399621327718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,power_law_1.2,0.04547199904918671
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,256,balanced,0.24868800242741904
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,power_law_1.01,0.12721279859542847
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,256,balanced,0.2686026692390442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,power_law_1.2,0.04580479860305786
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,power_law_1.2,0.022681599855422972
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,256,balanced,0.28140799204508465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,power_law_1.2,0.033369600772857666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,power_law_1.01,0.1456447958946228
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,256,balanced,0.302511990070343
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,power_law_1.2,0.08910080194473266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,power_law_1.2,0.0226623997092247
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,power_law_1.01,0.17955199480056763
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,power_law_1.2,0.0421887993812561
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,power_law_1.2,0.09552000164985656
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,power_law_1.2,0.02309119999408722
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,power_law_1.01,0.20999040603637695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,power_law_1.2,0.09880959987640381
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,power_law_1.2,0.04333440065383911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,power_law_1.01,0.11649279594421387
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,power_law_1.2,0.10478719472885131
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,power_law_1.2,0.02426239997148514
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,power_law_1.2,0.043296000361442565
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,power_law_1.01,0.1268288016319275
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,power_law_1.2,0.09678080081939697
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,power_law_1.2,0.0474047988653183
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,power_law_1.01,0.14515199661254882
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,power_law_1.01,0.02414720058441162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,power_law_1.2,0.09965440034866332
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,power_law_1.2,0.047814399003982544
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,power_law_1.01,0.16429439783096314
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,power_law_1.01,0.025964799523353576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,power_law_1.2,0.08566399812698364
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,power_law_1.01,0.2058880090713501
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,power_law_1.2,0.09202560186386108
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,power_law_1.01,0.02744320034980774
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,power_law_1.01,0.2512255907058716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,power_law_1.2,0.09500160217285156
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,power_law_1.01,0.034560000896453856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,power_law_1.2,0.10069760084152221
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,power_law_1.01,0.33400959968566896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,power_law_1.2,0.10684159994125367
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,power_law_1.01,0.035820800065994265
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,power_law_1.01,0.41486082077026365
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,power_law_1.01,0.2750976085662842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,power_law_1.2,0.11185920238494873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,power_law_1.2,0.11169279813766479
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,power_law_1.01,0.037248000502586365
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,power_law_1.2,0.025740799307823182
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,power_law_1.2,0.11662080287933349
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,power_law_1.01,0.34478719234466554
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,power_law_1.01,0.5775360107421875
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,power_law_1.01,0.03768959939479828
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,power_law_1.2,0.12775039672851562
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,power_law_1.2,0.03290880024433136
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,power_law_1.01,0.47820158004760743
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,power_law_1.01,0.7434688091278077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,power_law_1.2,0.1446272015571594
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,power_law_1.2,0.0342848002910614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,power_law_1.2,0.1677440047264099
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,power_law_1.01,0.6111807823181152
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,power_law_1.2,0.035743999481201175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,power_law_1.2,0.20872960090637208
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,power_law_1.01,1.0705984115600586
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,power_law_1.2,0.036575999855995175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,power_law_1.01,0.8617088317871093
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,power_law_1.2,0.0369024008512497
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,256,balanced,0.3282666603724162
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,power_law_1.01,1.390681552886963
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,power_law_1.2,0.040703999996185306
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,power_law_1.2,0.12319359779357911
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,256,balanced,0.3553813298543294
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,256,balanced,0.03738133360942205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,power_law_1.01,1.1262528419494628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,256,balanced,0.40143998463948566
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,power_law_1.2,0.03619199991226196
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,256,balanced,0.44908801714579266
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,power_law_1.01,0.03825919926166534
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,power_law_1.2,0.03696640133857727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,256,balanced,0.5212639967600504
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,256,balanced,0.6587839921315511
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,power_law_1.2,0.03683840036392212
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,power_law_1.01,1.6480960845947266
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,256,balanced,0.7985280354817709
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,power_law_1.2,0.03815680146217346
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,power_law_1.2,0.03875199854373932
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,256,balanced,1.0950240294138591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,power_law_1.2,0.25734400749206543
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,power_law_1.2,0.04151679873466492
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,256,balanced,1.3745919863382976
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,power_law_1.2,0.347871994972229
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,power_law_1.2,0.04471679925918579
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,power_law_1.01,2.1617984771728516
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,256,balanced,0.051957334081331887
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,256,balanced,0.05193066596984863
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,256,balanced,0.03770666569471359
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,256,balanced,0.052282666166623436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,power_law_1.2,0.4314176082611084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,power_law_1.2,0.13859200477600098
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,256,balanced,0.06018133461475372
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,256,balanced,1.9448000590006511
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,256,balanced,0.06459733347098033
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,power_law_1.2,0.10339839458465576
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,256,balanced,0.0819413314263026
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,power_law_1.2,0.15811840295791627
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,256,balanced,0.08990400036176045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,power_law_1.2,0.6091584205627442
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,256,balanced,0.09204266468683879
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,power_law_1.01,0.040659201145172116
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,256,balanced,0.2624800006548564
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,power_law_1.2,0.19969919919967652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,256,balanced,0.269050657749176
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,256,balanced,2.504624048868815
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,power_law_1.01,0.03798399865627289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,256,balanced,0.28516266743342084
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,power_law_1.2,0.7707327842712403
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,power_law_1.2,0.24676480293273925
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,256,balanced,0.29791466395060223
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,power_law_1.01,0.03854719996452331
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,power_law_1.2,0.3295167922973633
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,power_law_1.01,0.03790720105171204
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,power_law_1.2,1.1372032165527344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,power_law_1.2,0.41303038597106934
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,power_law_1.01,0.03825919926166534
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,power_law_1.2,0.04798080027103424
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,256,balanced,3.6914825439453125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,256,balanced,0.039477333426475525
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,256,balanced,0.04969066878159841
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,256,balanced,0.05816000203291575
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,power_law_1.01,0.03894400000572205
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,power_law_1.2,0.0585536003112793
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,256,balanced,0.07258666555086772
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,256,balanced,0.08277333279450734
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,power_law_1.2,1.4863231658935547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,256,balanced,0.08972799777984619
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,power_law_1.2,0.11193599700927734
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,power_law_1.01,0.04082559943199158
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,256,balanced,0.08275733391443889
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,power_law_1.2,0.06968960165977478
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,256,balanced,0.08780800302823384
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,power_law_1.2,0.12137600183486938
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,256,power_law_1.01,0.05648639798164368
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,power_law_1.01,0.044889599084854126
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,power_law_1.2,0.09354239702224731
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,power_law_1.2,0.12732160091400146
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,power_law_1.01,0.04728319942951202
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,power_law_1.2,0.12227200269699097
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,power_law_1.2,0.14663679599761964
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,256,balanced,0.31649067004521686
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,power_law_1.01,0.055776000022888184
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,256,balanced,4.830853462219238
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,256,balanced,0.3447839816411336
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,power_law_1.2,0.17872639894485473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,power_law_1.2,0.1819264054298401
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,256,balanced,0.375493327776591
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,power_law_1.01,0.06554880142211914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,256,balanced,0.40863998730977374
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,256,power_law_1.01,0.04154239892959595
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,power_law_1.2,0.23322880268096924
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,power_law_1.01,0.08675199747085571
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,power_law_1.2,0.5781568050384521
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,power_law_1.2,0.3401407957077026
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,power_law_1.01,0.1107200026512146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,256,power_law_1.01,0.042393600940704344
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,power_law_1.2,0.7499711990356446
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,power_law_1.01,0.16238080263137816
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,power_law_1.2,0.4459263801574707
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,256,power_law_1.01,0.043961599469184875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,256,balanced,0.2528533339500427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,256,balanced,0.27404266595840454
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,power_law_1.01,0.21022720336914064
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,256,balanced,0.28756266832351685
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,power_law_1.2,1.0760255813598634
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,256,power_law_1.01,0.047366398572921756
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,256,power_law_1.01,0.054527997970581055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,256,power_law_1.01,0.0551360011100769
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,power_law_1.01,0.29774720668792726
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,256,power_law_1.01,0.056524801254272464
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,256,power_law_1.01,0.05560960173606873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,256,power_law_1.01,0.05480319857597351
VLLM,0.17.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,power_law_1.01,0.3926912069320679
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,power_law_1.2,0.21596159934997558
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,power_law_1.2,1.3998144149780274
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,256,power_law_1.01,0.07227519750595093
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,256,balanced,0.4782293240229289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,256,power_law_1.01,0.0579584002494812
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,256,power_law_1.01,0.05542399883270264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,256,power_law_1.2,0.05668479800224304
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,256,balanced,0.6025439898173014
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,256,power_law_1.01,0.08158079981803894
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,256,power_law_1.2,0.04190079867839813
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,256,power_law_1.01,0.08095359802246094
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,256,balanced,0.7261066436767578
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,256,power_law_1.01,0.05818880200386047
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,256,power_law_1.2,0.055846399068832396
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,256,power_law_1.01,0.08836479783058167
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,256,power_law_1.2,0.04256640076637268
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,256,power_law_1.01,0.105132794380188
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,256,power_law_1.01,0.061478400230407716
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,256,power_law_1.01,0.08216320276260376
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,256,power_law_1.2,0.05612159967422485
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,256,balanced,0.30806400378545123
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,256,power_law_1.01,0.09653120040893555
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,256,power_law_1.01,0.08763520121574402
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,256,power_law_1.01,0.07781760096549988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,256,power_law_1.2,0.05861120223999024
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,256,power_law_1.01,0.08638719916343689
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,256,power_law_1.2,0.08138239979743958
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,256,power_law_1.01,0.09291520118713378
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,power_law_1.2,0.2905535936355591
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,256,power_law_1.2,0.10536320209503174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,256,power_law_1.01,0.26049280166625977
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,power_law_1.2,0.3630016088485718
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,256,power_law_1.2,0.0967423975467682
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,256,power_law_1.01,0.2678015947341919
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,256,balanced,0.9859253565470377
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,power_law_1.2,0.5162943840026856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,256,power_law_1.2,0.10080000162124633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,256,power_law_1.01,0.2830976009368896
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,256,power_law_1.2,0.0436928004026413
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,256,balanced,1.229967991511027
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,256,power_law_1.2,0.10149120092391968
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,power_law_1.2,0.6459712028503418
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,256,power_law_1.01,0.10024960041046142
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,256,power_law_1.01,0.29776639938354493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,256,power_law_1.2,0.046937599778175354
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,256,power_law_1.01,0.25265278816223147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,256,power_law_1.2,0.10536320209503174
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,256,balanced,1.7341012954711914
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,256,power_law_1.01,0.10183680057525635
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,256,power_law_1.01,0.315500807762146
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,power_law_1.2,0.9332672119140625
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,256,power_law_1.01,0.27105278968811036
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,256,power_law_1.2,0.2476288080215454
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,256,power_law_1.2,0.056678402423858645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,256,power_law_1.01,0.10443520545959473
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,256,power_law_1.01,0.34501121044158933
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,256,power_law_1.2,0.2659904003143311
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,256,power_law_1.01,0.2473344087600708
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,256,balanced,2.2242026329040527
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,256,power_law_1.2,0.07215359807014465
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,power_law_1.2,1.228006362915039
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,256,power_law_1.01,0.382694411277771
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,256,power_law_1.2,0.2804287910461426
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,256,power_law_1.01,0.2664704084396362
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,256,power_law_1.2,0.08133119940757752
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,256,power_law_1.01,0.28022398948669436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,256,power_law_1.01,0.4218751907348633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,256,power_law_1.2,0.30524160861968996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,256,power_law_1.2,0.08634240031242371
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,256,power_law_1.01,0.3026367902755737
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,256,power_law_1.01,0.4887296199798584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,256,balanced,3.2286240259806314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,power_law_1.2,1.822003173828125
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,256,power_law_1.01,0.6317696094512939
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,256,power_law_1.01,0.28558080196380614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,power_law_1.2,2.398624038696289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,256,power_law_1.01,0.7757376194000244
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,256,balanced,0.3362133502960205
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,256,balanced,4.222037315368652
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,256,power_law_1.01,0.3080895900726318
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,256,balanced,0.3622986475626628
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,256,balanced,0.4177600145339966
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,256,power_law_1.2,0.05674239993095398
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,256,power_law_1.01,1.0423359870910645
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,256,balanced,0.4737813472747803
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,256,balanced,0.553114652633667
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,256,power_law_1.2,0.05653759837150574
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,256,power_law_1.2,0.0818880021572113
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,256,balanced,0.7126453717549642
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,256,power_law_1.01,1.3303296089172363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,256,power_law_1.2,0.0875711977481842
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,256,balanced,0.8694026470184326
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,256,power_law_1.2,0.2522752046585083
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,256,balanced,6.257749557495117
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,256,power_law_1.2,0.2710975885391235
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,256,power_law_1.01,1.8682880401611328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,256,power_law_1.2,0.2851968050003052
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,256,power_law_1.2,0.30962560176849363
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,256,power_law_1.01,2.4406208038330077
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,256,power_law_1.2,0.05666559934616089
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,256,power_law_1.2,0.3354624032974243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,256,power_law_1.2,0.05909759998321533
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,256,balanced,1.1928426424662273
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,256,balanced,8.250682830810547
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,256,power_law_1.2,0.3623296022415161
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,256,power_law_1.2,0.06490240097045899
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,256,balanced,1.5073013305664062
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,256,power_law_1.01,0.33166720867156985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,256,power_law_1.01,3.5307968139648436
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,256,power_law_1.2,0.41473917961120604
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,256,power_law_1.2,0.07791360020637512
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,256,power_law_1.01,0.36154239177703856
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,256,power_law_1.2,0.08658559918403626
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,256,power_law_1.2,0.4671807765960693
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,256,power_law_1.01,0.4100031852722168
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,256,power_law_1.2,0.0917952001094818
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,256,power_law_1.01,0.4625216007232666
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,256,power_law_1.01,0.3363584041595459
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,256,power_law_1.2,0.555731201171875
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,256,power_law_1.2,0.2596544027328491
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,256,power_law_1.01,0.3646591901779175
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,256,power_law_1.01,0.5410304069519043
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,256,power_law_1.2,0.718559980392456
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,256,power_law_1.2,0.26808319091796873
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,256,power_law_1.01,0.4204224109649658
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,256,power_law_1.01,0.7006336212158203
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,256,power_law_1.01,4.646681594848633
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,256,power_law_1.2,0.28222079277038575
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,256,power_law_1.2,0.8973376274108886
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,256,power_law_1.01,0.47800321578979493
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,256,power_law_1.2,0.2977792024612427
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,256,power_law_1.01,0.8556415557861328
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,256,power_law_1.01,0.5613120079040528
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,256,balanced,2.1462079683939614
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,256,power_law_1.2,0.31500160694122314
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,256,power_law_1.2,1.2024640083312987
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,256,power_law_1.01,1.1572287559509278
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,256,power_law_1.01,0.7290624141693115
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,256,power_law_1.2,0.34995839595794676
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,256,balanced,2.777263959248861
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,256,power_law_1.01,0.8986751556396484
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,256,power_law_1.2,1.536838436126709
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,256,power_law_1.2,0.38719360828399657
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,256,power_law_1.2,0.3370752096176147
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,256,power_law_1.01,1.4792192459106446
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,256,power_law_1.2,0.42423038482666015
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,256,power_law_1.2,0.3638144016265869
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,256,power_law_1.01,1.225990390777588
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,256,power_law_1.2,0.5008575916290283
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,256,power_law_1.2,2.231827163696289
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,256,power_law_1.2,0.422976016998291
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,256,power_law_1.01,2.1240703582763674
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,256,power_law_1.01,1.560915184020996
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,256,power_law_1.2,0.6462207794189453
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,256,power_law_1.2,0.481990385055542
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,256,power_law_1.2,0.5688447952270508
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,256,power_law_1.2,0.7852352142333985
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,256,power_law_1.2,2.975142478942871
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,256,power_law_1.01,2.7271488189697264
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,256,power_law_1.01,2.222431945800781
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,256,power_law_1.2,0.7383743762969971
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,256,power_law_1.2,1.077286434173584
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,256,power_law_1.01,6.882303619384766
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,256,power_law_1.2,0.9094464302062988
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,256,balanced,4.0948746999104815
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,256,power_law_1.2,1.4065792083740234
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,256,power_law_1.2,1.2612671852111816
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,256,power_law_1.01,2.8908031463623045
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,256,power_law_1.01,4.103833770751953
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,256,power_law_1.2,4.448153686523438
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,256,power_law_1.2,1.9416959762573243
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,256,power_law_1.2,1.5994303703308106
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,256,balanced,5.36837895711263
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,256,power_law_1.2,2.5364095687866213
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,256,power_law_1.01,4.242227172851562
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,256,power_law_1.2,2.278361511230469
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,256,power_law_1.01,9.071315002441406
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,256,power_law_1.01,5.324121475219727
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,256,power_law_1.2,3.6710399627685546
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,256,power_law_1.2,2.967744064331055
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,256,power_law_1.2,5.714425659179687
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,256,power_law_1.01,5.599577713012695
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,256,power_law_1.2,4.94890251159668
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,256,power_law_1.2,7.263033294677735
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,256,power_law_1.2,4.392966461181641
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,256,power_law_1.2,9.77017593383789
VLLM,0.17.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,256,power_law_1.2,5.788345718383789
