framework,version,device,op_name,kernel_source,m,k,quant_dtype,latency
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,16384,fp8,0.2584522724151611
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,12288,fp8,0.1946069399515788
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,10240,fp8,0.1627295970916748
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,8192,fp8,0.1308842658996582
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,7168,fp8,0.11493013699849446
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,6144,fp8,0.09880853494008382
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,5120,fp8,0.08287680149078369
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,4096,fp8,0.06724586486816406
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,3584,fp8,0.05937920014063517
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,3072,fp8,0.05107306639353434
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,2560,fp8,0.04331093231836955
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,2048,fp8,0.03297599951426188
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,1536,fp8,0.01853546698888143
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,1024,fp8,0.012818132837613424
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,768,fp8,0.010041600465774536
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,512,fp8,0.007451733450094859
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,256,fp8,0.0046847999095916745
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,128,fp8,0.0037119999527931214
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,64,fp8,0.00317546675602595
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,32,fp8,0.0030303999781608583
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8192,65536,fp8,1.0352511723836264
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,16384,fp8,0.13082880179087322
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,12288,fp8,0.09909013112386068
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,10240,fp8,0.08302720387776694
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,8192,fp8,0.06701119740804037
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,7168,fp8,0.05910506645838419
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,6144,fp8,0.051132798194885254
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,5120,fp8,0.04350080092748006
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,4096,fp8,0.03234346707661946
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,3584,fp8,0.023374933004379272
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,3072,fp8,0.018372267484664917
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,2560,fp8,0.015366400281588236
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,2048,fp8,0.012822399536768595
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,1536,fp8,0.010045866171518963
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,1024,fp8,0.007439999779065449
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,768,fp8,0.006001066664854685
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,512,fp8,0.004653866589069367
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,256,fp8,0.0037087999284267426
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,128,fp8,0.0031530665854612983
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,64,fp8,0.0030048000315825146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,32,fp8,0.0028042666614055633
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4096,65536,fp8,0.5158634821573893
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,16384,fp8,0.06710506280263265
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,12288,fp8,0.0514794667561849
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,10240,fp8,0.043347199757893876
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,8192,fp8,0.03334826628367106
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,7168,fp8,0.022422399123509726
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,6144,fp8,0.018263467152913413
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,5120,fp8,0.015367466211318969
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,4096,fp8,0.012784000237782797
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,3584,fp8,0.011352533102035522
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,3072,fp8,0.01014293332894643
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,2560,fp8,0.00878613293170929
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,2048,fp8,0.007331199944019318
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,1536,fp8,0.0060703997810681665
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,1024,fp8,0.004644266764322917
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,768,fp8,0.00408746674656868
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,512,fp8,0.0037109332780043284
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,256,fp8,0.003151999910672506
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,128,fp8,0.0030048000315825146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,64,fp8,0.0028053333361943563
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,32,fp8,0.002674133330583572
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2048,65536,fp8,0.2581962744394938
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,16384,fp8,0.032015999158223465
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,12288,fp8,0.018329600493113198
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,10240,fp8,0.01548906664053599
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,8192,fp8,0.01269760032494863
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,7168,fp8,0.0114656001329422
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,6144,fp8,0.010130133231480916
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,5120,fp8,0.008694400389989216
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,4096,fp8,0.007427200178305308
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,3584,fp8,0.006680533289909363
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,3072,fp8,0.005992533266544342
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,2560,fp8,0.005406933526198069
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,2048,fp8,0.004655999938646952
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,1536,fp8,0.004196266829967499
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,1024,fp8,0.0037109332780043284
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,768,fp8,0.003390933324893316
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,512,fp8,0.00325546662012736
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,256,fp8,0.0029098667204380036
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,128,fp8,0.002799999962250392
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,64,fp8,0.002661333233118057
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,32,fp8,0.002755200117826462
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1024,65536,fp8,0.13065600395202637
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,16384,fp8,0.01812160015106201
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,12288,fp8,0.014099199573198953
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,10240,fp8,0.012008532881736755
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,8192,fp8,0.010115200281143188
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,7168,fp8,0.009089066584904989
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,6144,fp8,0.007982933521270752
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,5120,fp8,0.007089066505432129
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,4096,fp8,0.005996799965699514
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,3584,fp8,0.005409066875775656
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,3072,fp8,0.005051733553409576
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,2560,fp8,0.004546133180459341
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,2048,fp8,0.00418453315893809
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,1536,fp8,0.003818666686614355
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,1024,fp8,0.0034944000343481696
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,768,fp8,0.0032437334458033243
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,512,fp8,0.0031456001102924346
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,256,fp8,0.002872533351182938
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,128,fp8,0.0027434666951497394
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,64,fp8,0.0027818667391935987
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,32,fp8,0.002669866631428401
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,768,65536,fp8,0.09912853240966797
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,16384,fp8,0.012846933801968894
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,12288,fp8,0.010026666522026061
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,10240,fp8,0.008775466680526733
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,8192,fp8,0.007437866429487865
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,7168,fp8,0.006679466863473256
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,6144,fp8,0.0060928001999855045
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,5120,fp8,0.005329066514968872
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,4096,fp8,0.0047765334447224935
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,3584,fp8,0.004349866509437561
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,3072,fp8,0.0041909332076708475
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,2560,fp8,0.0038880000511805216
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,2048,fp8,0.0036373332142829893
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,1536,fp8,0.003489066660404205
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,1024,fp8,0.0031989333530267083
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,768,fp8,0.0031466667850812277
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,512,fp8,0.0030218665798505146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,256,fp8,0.0027434666951497394
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,128,fp8,0.0027882667879263563
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,64,fp8,0.0026858667532602947
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,32,fp8,0.002666666607062022
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,512,65536,fp8,0.06708799997965495
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,16384,fp8,0.010011733571688334
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,12288,fp8,0.008099199831485748
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,10240,fp8,0.007108266651630402
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,8192,fp8,0.005980800092220307
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,7168,fp8,0.005527466535568237
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,6144,fp8,0.004973866542180379
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,5120,fp8,0.00454720010360082
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,4096,fp8,0.0041802664597829185
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,3584,fp8,0.003969066590070724
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,3072,fp8,0.003918933371702829
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,2560,fp8,0.0037162666519482933
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,2048,fp8,0.003390933324893316
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,1536,fp8,0.003337600082159042
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,1024,fp8,0.003067733347415924
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,768,fp8,0.0029472000896930695
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,512,fp8,0.002844800055027008
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,256,fp8,0.0027029333015282948
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,128,fp8,0.00276053324341774
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,64,fp8,0.0026581334571043652
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,32,fp8,0.0027488000690937043
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,384,65536,fp8,0.05121386845906576
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,16384,fp8,0.00746666689713796
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,12288,fp8,0.006123733520507812
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,10240,fp8,0.005356800059477488
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,8192,fp8,0.00476800004641215
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,7168,fp8,0.004347733159859975
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,6144,fp8,0.00409706657131513
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,5120,fp8,0.003970133264859518
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,4096,fp8,0.0036309334139029183
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,3584,fp8,0.0036447999378045404
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,3072,fp8,0.00347626656293869
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,2560,fp8,0.003302400062481562
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,2048,fp8,0.0032629333436489105
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,1536,fp8,0.003048533449570338
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,1024,fp8,0.0030165334542592366
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,768,fp8,0.0027701333165168762
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,512,fp8,0.002808533360560735
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,256,fp8,0.0027242665489514667
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,128,fp8,0.0026869334280490874
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,64,fp8,0.0027594665686289472
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,32,fp8,0.002665599932273229
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,256,65536,fp8,0.032893866300582886
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,16384,fp8,0.00609493354956309
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,12288,fp8,0.005022933085759481
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,10240,fp8,0.004645333190759023
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,8192,fp8,0.004120533416668574
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,7168,fp8,0.003999999910593033
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,6144,fp8,0.003914666672547659
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,5120,fp8,0.00365226666132609
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,4096,fp8,0.0035018667578697203
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,3584,fp8,0.003386666625738144
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,3072,fp8,0.0033557333052158357
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,2560,fp8,0.003172266731659571
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,2048,fp8,0.00315733328461647
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,1536,fp8,0.003052799900372823
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,1024,fp8,0.002773333340883255
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,768,fp8,0.002808533360560735
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,512,fp8,0.00271573339899381
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,256,fp8,0.002703999976317088
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,128,fp8,0.00275093341867129
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,64,fp8,0.0026880001028378804
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,32,fp8,0.0027615999182065325
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,192,65536,fp8,0.018222934007644652
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,16384,fp8,0.0053269331653912864
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,12288,fp8,0.0046528001626332605
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,10240,fp8,0.004227200150489807
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,8192,fp8,0.003870933254559835
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,7168,fp8,0.003866666555404663
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,6144,fp8,0.003618133316437403
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,5120,fp8,0.0035434665779272715
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,4096,fp8,0.003419733295838038
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,3584,fp8,0.0032170665760835014
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,3072,fp8,0.003257599969704946
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,2560,fp8,0.003107200066248576
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,2048,fp8,0.0029887999097506206
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,1536,fp8,0.002963199963172277
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,1024,fp8,0.0027317332724730173
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,768,fp8,0.0028149334092934927
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,512,fp8,0.002775466690460841
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,256,fp8,0.0026933332284291584
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,128,fp8,0.002755200117826462
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,64,fp8,0.002647466709216436
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,32,fp8,0.002754133443037669
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,160,65536,fp8,0.01539413332939148
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,16384,fp8,0.004775466521581014
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,12288,fp8,0.004110933343569437
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,10240,fp8,0.0039007999002933502
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,8192,fp8,0.0036949334045251214
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,7168,fp8,0.0035743998984495797
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,6144,fp8,0.0034986667335033415
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,5120,fp8,0.0034058667719364167
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,4096,fp8,0.0031658666829268134
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,3584,fp8,0.0032170665760835014
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,3072,fp8,0.003070933371782303
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,2560,fp8,0.003005866706371307
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,2048,fp8,0.003013333429892858
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,1536,fp8,0.002771199991305669
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,1024,fp8,0.0028064000109831494
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,768,fp8,0.002705066651105881
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,512,fp8,0.00276053324341774
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,256,fp8,0.0026581334571043652
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,128,fp8,0.002733866622050603
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,64,fp8,0.002717866748571396
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,32,fp8,0.0026517334083716077
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,128,65536,fp8,0.012805333733558655
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,16384,fp8,0.004082133372624716
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,12288,fp8,0.0038133333126703895
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,10240,fp8,0.003702399879693985
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,8192,fp8,0.003402666747570038
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,7168,fp8,0.003436800092458725
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,6144,fp8,0.0033418667813142145
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,5120,fp8,0.0031466667850812277
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,4096,fp8,0.003171200056870779
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,3584,fp8,0.0029845332105954488
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,3072,fp8,0.003052799900372823
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,2560,fp8,0.002863999952872594
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,2048,fp8,0.002845866729815801
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,1536,fp8,0.002705066651105881
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,1024,fp8,0.0027114666998386385
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,768,fp8,0.0027872001131375628
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,512,fp8,0.0026880001028378804
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,256,fp8,0.0027615999182065325
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,128,fp8,0.0027413333455721537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,64,fp8,0.0026357332865397137
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,32,fp8,0.002739199995994568
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,96,65536,fp8,0.010034132997194927
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,16384,fp8,0.0038592000802357995
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,12288,fp8,0.0036906667053699495
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,10240,fp8,0.003419733295838038
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,8192,fp8,0.003403733422358831
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,7168,fp8,0.003209600100914637
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,6144,fp8,0.0032287999987602235
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,5120,fp8,0.00306986669699351
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,4096,fp8,0.0030805334448814393
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,3584,fp8,0.003031466652949651
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,3072,fp8,0.0028618666032950084
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,2560,fp8,0.002846933404604594
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,2048,fp8,0.00271573339899381
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,1536,fp8,0.002700799951950709
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,1024,fp8,0.0027872001131375628
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,768,fp8,0.0026677332818508146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,512,fp8,0.0027647999425729113
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,256,fp8,0.0027402666707833606
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,128,fp8,0.0026357332865397137
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,64,fp8,0.002735999971628189
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,32,fp8,0.0026410666604836782
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,80,65536,fp8,0.008770133058230083
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,16384,fp8,0.0036138666172822317
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,12288,fp8,0.00347626656293869
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,10240,fp8,0.003310933212439219
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,8192,fp8,0.00315733328461647
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,7168,fp8,0.0032106667757034303
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,6144,fp8,0.003044266750415166
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,5120,fp8,0.0030762667457262674
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,4096,fp8,0.003011200080315272
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,3584,fp8,0.0027903998891512555
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,3072,fp8,0.002842666705449422
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,2560,fp8,0.0027402666707833606
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,2048,fp8,0.0027072000006834666
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,1536,fp8,0.002796799937884013
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,1024,fp8,0.002679466704527537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,768,fp8,0.0027797333896160126
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,512,fp8,0.0026805333793163298
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,256,fp8,0.0027583998938401537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,128,fp8,0.002657066782315572
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,64,fp8,0.00275093341867129
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,32,fp8,0.002752000093460083
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,64,65536,fp8,0.007386666536331177
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,16384,fp8,0.003516799956560135
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,12288,fp8,0.003256533294916153
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,10240,fp8,0.0031626666585604347
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,8192,fp8,0.003151999910672506
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,7168,fp8,0.003032533327738444
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,6144,fp8,0.003045333425203959
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,5120,fp8,0.002951466788848241
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,4096,fp8,0.00276799996693929
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,3584,fp8,0.0028160000840822858
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,3072,fp8,0.0027274665733178455
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,2560,fp8,0.0028181334336598715
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,2048,fp8,0.0027274665733178455
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,1536,fp8,0.002776533365249634
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,1024,fp8,0.002699733277161916
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,768,fp8,0.0026890667776266735
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,512,fp8,0.002765866617361705
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,256,fp8,0.0026687999566396077
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,128,fp8,0.0027424000203609467
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,64,fp8,0.0027488000690937043
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,32,fp8,0.0026528000831604003
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,48,65536,fp8,0.006089599927266439
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,16384,fp8,0.00317546675602595
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,12288,fp8,0.0030613332986831666
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,10240,fp8,0.003089066594839096
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,8192,fp8,0.0029386666913827257
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,7168,fp8,0.002918400118748347
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,6144,fp8,0.0028714666763941447
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,5120,fp8,0.0027424000203609467
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,4096,fp8,0.0028192001084486645
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,3584,fp8,0.0027232001225153605
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,3072,fp8,0.0027200000981489818
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,2560,fp8,0.0027903998891512555
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,2048,fp8,0.002703999976317088
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,1536,fp8,0.002771199991305669
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,1024,fp8,0.0027637332677841187
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,768,fp8,0.002661333233118057
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,512,fp8,0.0027562665442625684
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,256,fp8,0.0026528000831604003
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,128,fp8,0.0026581334571043652
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,64,fp8,0.002746666719516118
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,32,fp8,0.002657066782315572
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,32,65536,fp8,0.004756266872088114
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,16384,fp8,0.002926933268706004
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,12288,fp8,0.0028736000259717304
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,10240,fp8,0.0027701333165168762
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,8192,fp8,0.0028010666370391845
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,7168,fp8,0.0028042666614055633
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,6144,fp8,0.0027189334233601887
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,5120,fp8,0.0027872001131375628
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,4096,fp8,0.0027104000250498454
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,3584,fp8,0.002705066651105881
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,3072,fp8,0.0027637332677841187
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,2560,fp8,0.0026943999032179515
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,2048,fp8,0.0027637332677841187
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,1536,fp8,0.0027669332921504974
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,1024,fp8,0.002656000107526779
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,768,fp8,0.0027413333455721537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,512,fp8,0.00264533335963885
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,256,fp8,0.002738133321205775
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,128,fp8,0.002665599932273229
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,64,fp8,0.002735999971628189
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,32,fp8,0.0026367999613285064
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,16,65536,fp8,0.0036149332920710243
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,16384,fp8,0.0027978666126728057
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,12288,fp8,0.0027146667242050173
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,10240,fp8,0.002784000088771184
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,8192,fp8,0.0027637332677841187
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,7168,fp8,0.0026869334280490874
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,6144,fp8,0.002771199991305669
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,5120,fp8,0.002665599932273229
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,4096,fp8,0.002647466709216436
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,3584,fp8,0.0027583998938401537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,3072,fp8,0.0026634665826956432
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,2560,fp8,0.0027445333699385325
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,2048,fp8,0.0026389333109060925
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,1536,fp8,0.0027413333455721537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,1024,fp8,0.00264533335963885
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,768,fp8,0.0027285332481066385
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,512,fp8,0.0027274665733178455
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,256,fp8,0.002632533262173335
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,128,fp8,0.0027413333455721537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,64,fp8,0.0026335999369621276
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,32,fp8,0.0026410666604836782
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,8,65536,fp8,0.003219199925661087
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,16384,fp8,0.002674133330583572
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,12288,fp8,0.002776533365249634
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,10240,fp8,0.0027594665686289472
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,8192,fp8,0.0026506667335828146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,7168,fp8,0.00276799996693929
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,6144,fp8,0.0026730666557947796
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,5120,fp8,0.0026634665826956432
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,4096,fp8,0.0027488000690937043
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,3584,fp8,0.002657066782315572
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,3072,fp8,0.0027530667682488757
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,2560,fp8,0.002752000093460083
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,2048,fp8,0.0026506667335828146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,1536,fp8,0.0027583998938401537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,1024,fp8,0.0026730666557947796
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,768,fp8,0.0026506667335828146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,512,fp8,0.002739199995994568
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,256,fp8,0.0026506667335828146
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,128,fp8,0.002739199995994568
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,64,fp8,0.0027327999472618104
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,32,fp8,0.0026549334327379864
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,4,65536,fp8,0.003019733230272929
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,16384,fp8,0.0026858667532602947
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,12288,fp8,0.0027615999182065325
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,10240,fp8,0.0026730666557947796
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,8192,fp8,0.0027530667682488757
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,7168,fp8,0.0026591998835404714
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,6144,fp8,0.002664533257484436
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,5120,fp8,0.0027402666707833606
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,4096,fp8,0.0026517334083716077
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,3584,fp8,0.0027488000690937043
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,3072,fp8,0.002737066646416982
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,2560,fp8,0.002666666607062022
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,2048,fp8,0.0027456000447273255
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,1536,fp8,0.002665599932273229
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,1024,fp8,0.002648533384005229
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,768,fp8,0.0027434666951497394
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,512,fp8,0.00264533335963885
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,256,fp8,0.0027456000447273255
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,128,fp8,0.002628266563018163
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,64,fp8,0.0027242665489514667
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,32,fp8,0.00262719988822937
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,2,65536,fp8,0.002799999962250392
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,16384,fp8,0.002739199995994568
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,12288,fp8,0.002644266684850057
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,10240,fp8,0.002734933296839396
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,8192,fp8,0.0026378666361172995
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,7168,fp8,0.0026346666117509207
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,6144,fp8,0.00275093341867129
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,5120,fp8,0.002656000107526779
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,4096,fp8,0.0027274665733178455
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,3584,fp8,0.0027104000250498454
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,3072,fp8,0.002639999985694885
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,2560,fp8,0.0027424000203609467
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,2048,fp8,0.002629333237806956
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,1536,fp8,0.002735999971628189
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,1024,fp8,0.002631466587384542
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,768,fp8,0.0027306665976842242
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,512,fp8,0.0026389333109060925
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,256,fp8,0.002643200010061264
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,128,fp8,0.002737066646416982
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,64,fp8,0.0026346666117509207
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,32,fp8,0.0027413333455721537
TRTLLM,1.2.0rc5,NVIDIA H200,scale_matrix,torch_ops,1,65536,fp8,0.002776533365249634
