framework,version,device,op_name,kernel_source,gemm_dtype,m,n,k,latency
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,16384,9.56569112141927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,16384,18.35902303059896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,12288,13.44101765950521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,12288,7.6169474283854175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,10240,6.784190877278645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,10240,12.336871337890624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,8192,9.871410115559895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,8192,5.1992950439453125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,8192,2.709735361735026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,7168,8.273712158203125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,7168,4.459951782226563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,10240,3.4685526529947914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,6144,7.043619283040364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,6144,4.070018259684245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,12288,4.319843037923177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,5120,5.889815266927084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,5120,3.389898681640625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,6144,1.904098129272461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,4096,4.42873789469401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,4096,2.492107645670573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,7168,2.2891082763671875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,3584,3.8206921895345056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,3584,1.9496042887369793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,5120,1.5746964772542318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,3072,3.353166961669922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,3072,1.463647969563802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,16384,6.007864379882813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,2560,2.666548156738281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,2560,1.0351999918619792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,3584,0.9602197647094727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,2048,2.1636566162109374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,2048,1.0700672149658204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,3072,0.7353898366292317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,1536,1.6743914286295574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,1536,0.7259946823120117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,4096,0.9624287923177084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,1024,1.018245315551758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,1024,0.4964010556538899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,1024,0.3466613451639811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,768,0.47550185521443683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,768,0.4651167869567871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,2560,0.6442197163899739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,512,0.3442421277364095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,512,0.40589119593302414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,1536,0.45716479619344075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,2048,0.5474016189575195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,256,0.225819730758667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,256,0.21211627324422203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,128,0.19358612696329752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,128,0.18291519482930502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,128,0.18312640190124513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,64,0.16489706039428711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,64,0.18453866640726727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,65536,32,0.16802453994750977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,65536,32,0.18454079627990722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,256,0.20527572631835939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,65536,10.102504475911458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,768,0.297158400217692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,65536,512,0.2603552023569743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,16384,1.977609634399414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,65536,18.0669921875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,16384,4.323444112141927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,12288,1.7226165771484374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,12288,3.1444971720377604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,10240,3.151282246907552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,10240,1.4156490325927735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,10240,0.6626368204752604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,8192,2.3934399922688803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,8192,1.1144341786702474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,12288,0.9646965026855469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,7168,1.9887893676757813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,7168,0.9400874455769858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,16384,1.1125386555989585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,6144,1.572003173828125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,6144,0.7124586741129557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,8192,0.5184416135152181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,5120,1.3305408477783203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,5120,0.5779914855957031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,7168,0.4566357294718425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,4096,0.867744000752767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,4096,0.6071349461873372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,5120,0.32991040547688805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,3584,0.6114144007364909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,3584,0.4326976140340169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,4096,0.2701141357421875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,3072,0.42198721567789715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,6144,0.3863295873006185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,3072,0.3200426737467448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,2560,0.5201354662577311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,2560,0.3951605478922526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,3584,0.2420992056528727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,2048,0.28161067962646485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,2048,0.22643839518229164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,2560,0.18465813000996906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,1536,0.21753387451171874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,1536,0.1600106716156006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,3072,0.21079893112182618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,1024,0.14885652860005696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,1024,0.1199338674545288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,1024,0.09880212942759195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,768,0.11804800033569336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,768,0.10259520212809245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,1536,0.12937813599904377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,512,0.08923093477884927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,512,0.08335573673248291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,768,0.08438826402028402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,256,0.0599616010983785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,256,0.07030399640401205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,512,0.07418453693389893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,128,0.05189866622289022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,128,0.06444160143534342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,2048,0.1549418608347575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,64,0.04743146498998006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,64,0.062112001578013096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,16384,32,0.04680533409118652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,256,0.06226133505503336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,16384,32,0.060660266876220705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,128,0.056910932064056396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,65536,7.166983540852864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,16384,3.383001708984375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,65536,12.718802897135417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,16384,2.0001152038574217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,12288,2.50630620320638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,12288,1.2927903493245443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,12288,0.5915914535522461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,10240,2.157904052734375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,10240,1.0163285573323568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,16384,0.7754079818725585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,8192,1.399347178141276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,8192,0.8330218633015951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,10240,0.4999573389689128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,7168,0.9113952000935873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,7168,0.7893440246582031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,8192,0.3900992075602213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,6144,0.9807274500528971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,6144,0.6956000010172526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,7168,0.3404714584350586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,5120,0.7907648086547852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,5120,0.5255039850870769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,6144,0.30407466888427737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,4096,0.4349141438802083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,4096,0.3653696060180664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,5120,0.26126933097839355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,3584,0.41785599390665695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,3584,0.31638612747192385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,4096,0.21347413063049317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,3072,0.32816851933797203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,3072,0.32995840708414714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,3584,0.190227206548055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,2560,0.271617062886556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,2560,0.2372959931691488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,3072,0.17031680742899577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,2048,0.22064639727274576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,2048,0.1652234713236491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,2560,0.146834135055542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,1536,0.16739306449890137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,1536,0.12751466433207195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,1536,0.10200320084889729
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,1024,0.1149183988571167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,1024,0.0971722682317098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,2048,0.12400000095367432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,768,0.09002559979756673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,768,0.08472959995269776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,1024,0.07843733628590902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,512,0.06774400075276693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,512,0.06798293590545654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,768,0.06751253604888915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,256,0.046190933386484785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,256,0.06035946607589722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,256,0.050994133949279784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,128,0.0371338685353597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,128,0.05204693476359049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,128,0.046204801400502524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,64,0.03440213203430176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,64,0.05332906643549601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,12288,32,0.03742506504058838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,12288,32,0.05411306619644165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,512,0.059646932284037266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,65536,6.118403116861979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,65536,10.36548563639323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,16384,2.746337127685547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,16384,1.4080106099446614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,16384,0.6326741536458333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,12288,1.9041322072347004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,12288,1.0677696228027345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,16384,65536,7.474663289388021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,10240,1.4191893259684245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,10240,1.05513064066569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,12288,0.4811861356099446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,8192,0.7299797058105468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,8192,1.0117247899373374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,10240,0.40197013219197586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,7168,0.6248746871948242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,7168,0.649181874593099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,8192,0.3205087979634603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,6144,0.5434442520141601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,6144,0.45819733937581375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,7168,0.2851797421773275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,5120,0.44562772115071614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,5120,0.36375786463419596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,12288,65536,5.369616190592447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,4096,0.567955207824707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,4096,0.36857493718465173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,6144,0.25318293571472167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,3584,0.31660054524739584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,3584,0.25443946520487465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,5120,0.22050560315450035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,3072,0.26841920216878257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,3072,0.22053227424621583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,4096,0.18045760790506998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,2560,0.2291872024536133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,2560,0.18905599912007648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,3072,0.14230507214864094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,2048,0.18076586723327637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,2048,0.14042240778605145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,3584,0.16272212664286295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,1536,0.13959147135416666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,1536,0.1129034678141276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,2560,0.12568960189819336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,1024,0.0966912031173706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,1024,0.08665599822998046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,1536,0.08709759712219238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,768,0.07497599919637045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,768,0.07468693256378174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,2048,0.10436906814575195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,512,0.05624320109685262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,512,0.060078934828440345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,1024,0.06674133141835531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,256,0.040099199612935385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,256,0.0515936017036438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,768,0.057139201958974206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,128,0.03540906508763631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,128,0.0456000010172526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,128,0.04069439967473348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,64,0.02790186603864034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,64,0.044369065761566163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,10240,32,0.029521065950393676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,256,0.0441429336865743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,10240,32,0.045823999245961505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,512,0.05079466501871745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,65536,4.885655212402344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,16384,1.9882731119791668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,65536,8.553984578450521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,16384,1.2372117360432944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,12288,1.3528586069742838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,12288,1.0809504191080728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,12288,0.39711999893188477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,10240,0.7549919764200846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,10240,0.9042122522989908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,16384,0.526203727722168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,8192,0.9003189086914063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,8192,0.7288085301717122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,10240,0.3367818514506022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,7168,0.5542463938395182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,7168,0.6873258590698242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,8192,0.27001492182413733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,6144,0.4999050776163737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,6144,0.5751861572265625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,7168,0.24311787287394204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,5120,0.37184534072875974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,5120,0.3482975959777832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,10240,65536,4.263827260335287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,4096,0.5011413256327312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,4096,0.3458378791809082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,6144,0.21463467280069987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,3584,0.2593461354573568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,3584,0.20889172554016114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,4096,0.15192000071207684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,5120,0.19010772705078124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,3072,0.22210240364074707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,3072,0.1798378626505534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,2560,0.18019839922587078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,2560,0.15214187304178875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,3584,0.13855573336283367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,2048,0.14643413225809734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,2048,0.12738239765167236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,2560,0.10630613168080646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,1536,0.1117514689763387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,1536,0.10022613207499187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,3072,0.1204416036605835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,1024,0.07827946345011393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,1024,0.07491733233133951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,2048,0.08912853399912515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,768,0.061877334117889406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,768,0.06572266817092895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,768,0.04869759877522786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,512,0.046583465735117596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,512,0.05272639989852905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,1024,0.05654933452606201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,256,0.03205866614977519
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,256,0.045424000422159834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,1536,0.07386346658070883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,128,0.02869759996732076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,128,0.041725865999857586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,256,0.03869866530100505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,64,0.02332373261451721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,64,0.03948800166447957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,8192,32,0.02374826669692993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,8192,32,0.038423466682434085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,512,0.0436245322227478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,128,0.03580373525619507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,65536,4.223173268636068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,16384,1.808713658650716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,65536,7.3018442789713545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,16384,1.092748769124349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,12288,0.9310826619466146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,12288,0.9195583979288736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,12288,0.344596258799235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,10240,0.8544789632161459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,10240,0.7588554382324219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,16384,0.459775988260905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,8192,0.7395765304565429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,8192,0.5809386571248372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,10240,0.29591894149780273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,7168,0.4344053268432617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,7168,0.5091210683186849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,8192,0.24005653063456217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,6144,0.3688576062520345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,6144,0.3368149439493815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,7168,0.2155615965525309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,5120,0.3079061190287272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,5120,0.2796565373738607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,6144,0.19058666229248047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,4096,0.245576540629069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,4096,0.22538026173909506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,5120,0.1701855977376302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,3584,0.21555840174357094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,3584,0.18977492650349934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,4096,0.1360970656077067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,3072,0.18605119387308758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,3072,0.1630751927693685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,3584,0.12308906714121501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,2560,0.15624106725056966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,2560,0.13908373514811198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,3072,0.1077130635579427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,2048,0.12671786944071453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,2048,0.11419839859008789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,2560,0.09445013205210367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,1536,0.09699626763661703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,1536,0.09038613637288412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,2048,0.07912000020345053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,1024,0.0680074691772461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,1024,0.06789866288503012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,1536,0.06575466791788737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,768,0.05335466861724854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,768,0.060483201344807946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,1024,0.050556798775990806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,512,0.04013760089874267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,768,0.04341226816177368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,512,0.049148801962534586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,256,0.02802773316701253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,256,0.04244053363800049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,256,0.035358933607737224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,128,0.024567466974258424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,128,0.037946665287017824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,128,0.03256746729214986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,512,0.0387605349222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,64,0.020820266008377074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,7168,32,0.021091200908025107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,64,0.037622400124867755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,7168,32,0.03511573473612468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,65536,3.5406089782714845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,8192,65536,3.5362388610839846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,65536,6.22269541422526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,16384,1.3055615743001303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,16384,0.8890730539957682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,12288,0.783359972635905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,12288,0.7622602462768555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,12288,0.3192426681518555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,10240,0.5529674530029297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,10240,0.5489664077758789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,16384,0.41756480534871415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,8192,0.4370261192321777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,8192,0.46824105580647785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,10240,0.2751530647277832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,7168,0.38827625910441077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,7168,0.44991146723429365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,8192,0.22033920288085937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,6144,0.3279146512349447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,6144,0.32206719716389975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,7168,0.20037439664204917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,5120,0.2765781402587891
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,5120,0.2637269337972005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,6144,0.17677653630574544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,4096,0.21773120562235512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,4096,0.21188586552937827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,5120,0.1577013333638509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,3584,0.1934069315592448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,3584,0.17988160451253254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,4096,0.12561386426289875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,3072,0.1645066738128662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,3072,0.15158613522847492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,3072,0.10038932959238689
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,3584,0.11503466765085857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,2560,0.1382197380065918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,2560,0.12991039752960204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,2048,0.11299093564351399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,2048,0.10668266614278157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,2048,0.07333652973175049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,1536,0.08633920351664225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,2560,0.08740586439768473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,1536,0.08319679896036783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,1024,0.0608351985613505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,1024,0.06325013240178426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,1536,0.0603925347328186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,768,0.047005867958068846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,768,0.05615573326746622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,1024,0.04734613498051961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,512,0.03444053332010905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,512,0.053591465950012206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,768,0.04035626649856568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,256,0.030349866549173994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,256,0.0452949325243632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,512,0.036610134442647296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,128,0.019857066869735717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,256,0.03300799926122029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,128,0.035189334551493326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,64,0.017831466595331826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,64,0.03341226577758789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,6144,32,0.01818559964497884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,7168,65536,2.7843488057454424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,128,0.030590933561325074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,6144,32,0.03250880042711894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,65536,3.2545430501302084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,16384,1.053001594543457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,65536,5.410698445638021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,16384,0.845257568359375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,12288,0.8480639775594077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,12288,0.7708063761393229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,12288,0.27013867696126304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,10240,0.6306336085001628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,10240,0.6770591735839844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,16384,0.35156265894571936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,8192,0.37278292973836263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,8192,0.43822078704833983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,10240,0.2331914742787679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,7168,0.31808532079060875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,7168,0.5098464012145996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,8192,0.1904245376586914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,6144,0.2783402760823568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,6144,0.3374677340189616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,7168,0.17028800646464032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,5120,0.2270869255065918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,5120,0.24482986132303872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,6144,0.15304640134175618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,4096,0.18476160367329914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,4096,0.1851626714070638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,5120,0.13586452802022297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,3584,0.15861120223999023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,3584,0.15948479970296223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,3584,0.09878506660461425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,3072,0.13657600084940594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,3072,0.13647252718607586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,4096,0.10953919887542725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,2560,0.11719146569569905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,2560,0.11601280371348063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,2560,0.0764245351155599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,2048,0.09467519919077555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,2048,0.09629973570505777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,3072,0.08664960066477458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,1536,0.07078826427459717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,1536,0.0756821314493815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,1536,0.05312000115712484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,1024,0.04972373247146607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,1024,0.05721706549326579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,2048,0.06380480130513509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,768,0.04055146773656209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,768,0.0516480008761088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,1024,0.041042133172353105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,512,0.02860693335533142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,512,0.042827733357747394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,768,0.03510293165842692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,256,0.021244800090789794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,512,0.0315391997496287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,256,0.03670186599095662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,128,0.017373865842819212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,128,0.0330186665058136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,256,0.029706666866938274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,64,0.015925332903862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,128,0.0276853342851003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,64,0.03097066680590312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,5120,32,0.016716800133387246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,5120,32,0.030269867181777953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,65536,2.748454284667969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,6144,65536,2.324017079671224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,65536,4.2344820658365885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,16384,0.71888427734375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,16384,0.7345930735270183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,12288,0.5395743687947591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,12288,0.5620629628499348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,12288,0.22578560511271156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,10240,0.3668160120646159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,10240,0.5812735875447591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,16384,0.29130560557047525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,8192,0.29064534505208334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,8192,0.5087583859761555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,10240,0.19721066157023112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,7168,0.2575551986694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,7168,0.354913075764974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,8192,0.15923306147257488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,6144,0.21797866821289064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,6144,0.2561898708343506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,7168,0.14389759699503582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,5120,0.18361493746439617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,5120,0.21252479553222656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,6144,0.12921066284179689
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,4096,0.1453802744547526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,4096,0.1704586664835612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,5120,0.11543893019358317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,3584,0.12928746541341146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,3584,0.14247147242228192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,4096,0.09184640248616537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,3072,0.10973546504974366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,3072,0.12109440167744953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,3584,0.08337066968282064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,2560,0.09370773633321126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,2560,0.10355733235677082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,3072,0.07318080266316732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,2048,0.0750378688176473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,2048,0.08515946865081787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,2560,0.0642741322517395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,1536,0.05841919978459677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,1536,0.06796586513519287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,1536,0.045371735095977785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,1024,0.04058880011240641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,1024,0.05159680048624674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,2048,0.05404586791992187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,768,0.031177600224812824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,768,0.04542719920476278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,1024,0.03435946702957153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,512,0.023412267367045082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,512,0.03712213436762492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,768,0.029002666473388672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,256,0.016662399967511496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,256,0.03149440089861552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,5120,65536,1.745025126139323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,128,0.020666666825612388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,128,0.043087999025980636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,256,0.025575466950734455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,64,0.0132832000652949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,64,0.02740266720453898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,512,0.02650986711184184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,4096,32,0.01402773360411326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,128,0.024357332785924276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,4096,32,0.026741333802541095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,65536,2.4430185953776045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,65536,3.570780690511068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,16384,0.5275978724161784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,16384,0.6293994903564453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,12288,0.36670185724894205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,12288,0.5316704114278157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,12288,0.20215253829956054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,10240,0.3166858673095703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,10240,0.4642229398091634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,16384,0.26029866536458335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,8192,0.2544586658477783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,8192,0.38918720881144203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,10240,0.17706987063090007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,7168,0.22209173838297525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,7168,0.27967252731323244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,8192,0.1427786668141683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,6144,0.1907360076904297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,6144,0.24101012547810874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,7168,0.13031040032704672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,5120,0.15810880661010743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,5120,0.19909440676371257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,6144,0.11634666919708252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,4096,0.12648746967315674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,4096,0.15385492642720539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,5120,0.10458239714304607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,3584,0.11057493686676026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,3584,0.13258346716562908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,4096,0.08400853474934897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,3072,0.09534400304158529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,3072,0.11361066500345866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,4096,65536,1.124840545654297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,2560,0.07963840166727701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,2560,0.104912002881368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,3584,0.07578986485799154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,2048,0.06625599861145019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,2048,0.08024746576944987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,3072,0.06718506813049316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,1536,0.05105813344319662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,1536,0.06350719928741455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,2560,0.05824640194574991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,1024,0.034745601812998454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,1024,0.04797120094299316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,2048,0.048827731609344484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,768,0.027185066541035967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,768,0.04221546649932861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,1536,0.04110186497370402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,512,0.020134399334589638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,512,0.03516586621602376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,1024,0.03100373347600301
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,256,0.014645333091417948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,256,0.028841600815455122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,768,0.026199465990066527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,128,0.013318399588267008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,512,0.024734934171040855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,128,0.027049599091211955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,256,0.024165334304173787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,128,0.023196800549825033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,64,0.011242666840553283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3584,32,0.011595732967058818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,64,0.025443200270334882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3584,32,0.024715733528137208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,65536,2.9966697692871094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,65536,2.3869247436523438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,16384,0.6434015909830729
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,16384,0.6113194783528646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,12288,0.4022229194641113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,12288,0.4910431861877441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,12288,0.20015254020690917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,10240,0.28797438939412434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,10240,0.5320010821024577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,16384,0.2609130700429281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,8192,0.23230506579081217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,8192,0.30781332651774085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,10240,0.174563201268514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,7168,0.19848640759785968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,7168,0.2880597432454427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,8192,0.14291733105977375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,6144,0.1713813304901123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,6144,0.23270293871561684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,7168,0.1280405362447103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,5120,0.141102933883667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,5120,0.18983359336853028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,6144,0.11641279856363933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,4096,0.1146997372309367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,4096,0.14804479281107585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,5120,0.10291093190511066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,3584,0.09993173281351725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,3584,0.12739413579305012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,3584,0.0745855967203776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,3072,0.0861407995223999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,4096,0.08485013643900553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,3072,0.10930453141530354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,2560,0.07280000050862631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,2560,0.09263573487599691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,2560,0.057426134745279946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,2048,0.05919466813405355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,2048,0.07639573415120443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,3072,0.06614293257395426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,1536,0.04610346555709839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,1536,0.0612501343091329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,2048,0.04767573277155558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,1536,0.03960213263829549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,1024,0.031436800956726074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,1024,0.04502933422724406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,768,0.025525333484013875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,768,0.0396448016166687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3584,65536,1.103271484375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,512,0.026629332701365156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,512,0.03291093309720357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,768,0.025967999299367266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,256,0.013398399949073792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,1024,0.0299072007338206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,256,0.02908373276392619
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,512,0.023991467555363973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,128,0.01204159955183665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,128,0.025270400444666545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,64,0.010146133104960124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,128,0.02267306645711263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,64,0.02493120034535726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,256,0.023541333278020223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,3072,32,0.010857599973678588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,3072,32,0.02290453314781189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,65536,2.183995819091797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,65536,2.926523844401042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,16384,0.36766398747762047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,16384,0.5651381174723308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,12288,0.27494932810465494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,12288,0.588749885559082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,12288,0.17820693651835123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,10240,0.2584341367085775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,10240,0.37784318923950194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,16384,0.23010133107503256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,8192,0.2086538632710775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,8192,0.29618988037109373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,10240,0.15501440366109212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,7168,0.17805226643880206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,7168,0.2596149285634359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,8192,0.12639573415120442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,6144,0.15350507100423177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,6144,0.22480533917744955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,7168,0.11400427023569744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,5120,0.1277120033899943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,5120,0.1834869384765625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,6144,0.1041098674138387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,4096,0.10439039866129558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,4096,0.14310293197631835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,5120,0.09380053679148356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,3584,0.09143679936726888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,3584,0.12147200107574463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,3584,0.06749333540598551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,3072,0.07975467046101889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,4096,0.07523840268452962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,3072,0.1038698673248291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,2560,0.06710506280263265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,2560,0.08868587017059326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,3072,0.05999146699905396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,2048,0.05534826517105103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,2048,0.07411306699117025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,2560,0.05151253143946329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,1536,0.0416159987449646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,1536,0.0574346661567688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,2048,0.043119998772939046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,1024,0.02938026587168376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,1024,0.04256213506062825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,1536,0.03497920036315918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,768,0.022658133506774904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,768,0.03713493347167969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,768,0.02267199953397115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,512,0.01660053332646688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,1024,0.026982400814692182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,512,0.03091520071029663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,256,0.011819733182589214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,256,0.026909865935643512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,256,0.021974400679270426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,128,0.009253333012262981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,128,0.023849600553512575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,512,0.02185813387235006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,64,0.008326399823029835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,64,0.02276159922281901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,128,0.02101226647694906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2560,32,0.008422399560610454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2560,32,0.022425599892934165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,3072,65536,1.070561091105143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,65536,1.6317354838053386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,16384,0.5036032040913899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,65536,2.0470218658447266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,16384,0.5340394973754883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,12288,0.22810559272766112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,12288,0.4062656084696452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,12288,0.15505599975585938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,10240,0.19991572697957355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,10240,0.33946345647176107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,16384,0.19841386477152506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,8192,0.15952000617980958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,8192,0.2906581242879232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,10240,0.13584319750467938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,7168,0.14101546605428059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,7168,0.24020053545633951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,8192,0.10997866789499919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,6144,0.12044800122578939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,6144,0.20236053466796874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,7168,0.10047787030537922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,5120,0.10115306377410889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,5120,0.16755733489990235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,6144,0.09095146656036376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,4096,0.08009493350982666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,4096,0.12744106451670328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,4096,0.0662496010462443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,3584,0.0713088035583496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,5120,0.08375146389007568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,3584,0.10919679800669353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,3072,0.0590666651725769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,3072,0.09285226662953695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,3072,0.05206079880396525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,2560,0.04965333143870036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,3584,0.06050026814142863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,2560,0.07934292952219645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,2048,0.0403658668200175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,2048,0.06368746757507324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,2048,0.03717120091120402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,1536,0.03049386739730835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,1536,0.049482667446136476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,2560,0.04583359956741333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,1024,0.02175040046374003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,1024,0.037251198291778566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2560,65536,0.8509951909383139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,768,0.017564799388249716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,768,0.044733866055806475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,1024,0.02403093377749125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,512,0.01286720037460327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,1536,0.030483200152715044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,512,0.029275733232498168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,256,0.009801600376764934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,768,0.020294400056203206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,256,0.026364799340565997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,128,0.008798933029174805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,128,0.022826667626698813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,256,0.020670932531356812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,64,0.007580799857775371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,512,0.019363200664520262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,128,0.01995733380317688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,2048,32,0.007575466732184092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,64,0.02176106572151184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,2048,32,0.022388267517089843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,65536,1.121170171101888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,16384,0.40451520284016923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,65536,1.9088874816894532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,16384,0.5106591860453288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,12288,0.17502719561258953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,12288,0.37210346857706705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,12288,0.13082559903462726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,10240,0.15029013951619466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,10240,0.3108511924743652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,16384,0.16832213401794432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,8192,0.12373653252919514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,8192,0.2516512076059977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,10240,0.11618133385976155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,7168,0.10585920015970866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,7168,0.21893866856892905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,7168,0.0861525297164917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,6144,0.09117013613382975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,6144,0.18802347183227539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,8192,0.09411946932474771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,5120,0.07775253454844157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,5120,0.15512213706970215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,5120,0.07255573272705078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,4096,0.06258453528086344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,6144,0.07898773352305094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,4096,0.11813013553619385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,3584,0.05480213165283203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,3584,0.10122666358947754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,3584,0.05139946540196737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,3072,0.04894293149312337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,3072,0.08626026312510172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,4096,0.05752426783243815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,2560,0.039274664719899495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,2560,0.07128213246663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,2560,0.03880426486333211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,2048,0.03157866597175598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,2048,0.057708799839019775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,3072,0.04583253463109334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,1536,0.02502400080362956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,1536,0.045407998561859134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,1536,0.026972800493240356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,1024,0.018658133347829182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,2048,0.03186986645062764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,1024,0.036551467577616376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,768,0.014988799889882406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,768,0.03368106683095296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,1024,0.020614399512608846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,512,0.011964799960454305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,512,0.028617600599924724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,768,0.01798186699549357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,256,0.010472533106803895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,256,0.025297067562739056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,512,0.017127466201782227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,128,0.00790293316046397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,128,0.02426240046819051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,256,0.018623999754587808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,64,0.008004266520341237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,128,0.018181333939234413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1536,32,0.006995200117429097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,64,0.020936532815297445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1536,32,0.020838399728139244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,2048,65536,0.7285120010375976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,65536,0.6093407948811849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,16384,0.15861120223999023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,16384,0.5317536036173502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,65536,1.7341419219970704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,12288,0.1163530667622884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,12288,0.3396661440531413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,12288,0.10781866709391277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,10240,0.10073386828104655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,10240,0.2839008013407389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,16384,0.13755946159362792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,8192,0.08411946296691894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,8192,0.22966292699178062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,10240,0.09597226778666178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,7168,0.07203413645426432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,7168,0.20014293988545737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,8192,0.07837759653727214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,6144,0.06319786707560221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,6144,0.17197972933451336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,7168,0.07216426531473795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,5120,0.05189333359400431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,5120,0.1401866594950358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,6144,0.06697706381479898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,4096,0.04262933333714803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,4096,0.10725226402282714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,5120,0.0623093326886495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,3584,0.036722131570180255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,3584,0.09012693564097086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,4096,0.049304533004760745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,3072,0.03254400094350179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,3072,0.07612480322519938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,3584,0.04409066836039226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,2560,0.026779733101526898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,2560,0.06112639904022217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,2560,0.032561065753300984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,2048,0.022408533096313476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,3072,0.038022398948669434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,2048,0.048493866125742593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,1536,0.01763946612675985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,1536,0.040046934286753336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,1536,0.023085866371790567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,1024,0.01239359974861145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,1024,0.03306666612625122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,2048,0.02726400097211202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,768,0.010493866602579753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,768,0.029174399375915528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,1024,0.01736533244450887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,512,0.008085333307584127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,512,0.024733867247899374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,768,0.015507200360298156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,256,0.006413866579532623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,512,0.014857600132624308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,256,0.022699733575185142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,256,0.01692906618118286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,128,0.005868799984455109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,128,0.02059946656227112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,128,0.016926934321721397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,64,0.0049898669123649595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,64,0.018833067019780478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,1024,32,0.0049674664934476215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,1024,32,0.019309866428375243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,65536,0.46240746180216474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1536,65536,0.6133632024129232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,65536,1.6568064371744793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,16384,0.12238506476084392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,16384,0.42692801157633464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,12288,0.09027413527170816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,12288,0.3247701327006022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,12288,0.10746880372365315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,10240,0.08394986788431803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,10240,0.2760693232218424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,16384,0.13654719988505046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,8192,0.06943786938985189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,8192,0.22117439905802408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,10240,0.09566933314005534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,7168,0.05913493235905966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,7168,0.19432320594787597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,8192,0.0774890661239624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,6144,0.05150933265686035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,6144,0.16632533073425293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,7168,0.07192320028940836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,5120,0.04227306842803955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,5120,0.13525333404541015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,6144,0.06672853628794352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,4096,0.03520960013071696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,4096,0.10181653499603271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,1024,65536,0.4957898775736491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,3584,0.030503465731938677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,3584,0.08601493040720622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,5120,0.061957331498463955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,3072,0.025989333788553875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,3072,0.07058666547139486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,4096,0.04802560011545817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,3584,0.04233920176823934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,2560,0.022344533602396646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,2560,0.05649493137995402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,2048,0.01846826672554016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,3072,0.03623146613438924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,2048,0.047305599848429365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,1536,0.01441493332386017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,1536,0.03843413194020589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,2560,0.031830400228500366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,1024,0.010674132903416952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,1536,0.022670932610829673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,2048,0.026829866568247478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,768,0.008676266670227051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,1024,0.030956800778706866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,768,0.028884265820185346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,512,0.00676800012588501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,512,0.02474453250567118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,1024,0.017417599757512413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,768,0.014973866939544677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,256,0.005941333373387655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,512,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,128,0.005100800096988678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,256,0.022957867383956908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,128,0.020637865861256918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,64,0.005052799979845682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,256,0.017012266318003337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,128,0.016899200280507405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,768,32,0.005120000243186951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,64,0.019008000691731773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,768,32,0.01767680048942566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,65536,0.32337493896484376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,16384,0.08525333404541016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,16384,0.4746933301289876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,65536,1.5786869049072265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,12288,0.06363093455632528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,12288,0.30832961400349934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,12288,0.08412480354309082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,10240,0.053741868336995444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,10240,0.2583338737487793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,16384,0.10673600037892658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,8192,0.04472106695175171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,8192,0.2082805315653483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,10240,0.07637653350830079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,7168,0.03890560070673625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,7168,0.1814677397410075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,8192,0.06291733185450235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,6144,0.03464320103327433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,6144,0.1578154722849528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,7168,0.058652798334757485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,5120,0.02872213323911031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,5120,0.1270944039026896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,6144,0.055270401636759434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,4096,0.02338026762008667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,4096,0.09475519657135009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,5120,0.05184640089670817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,3584,0.02039146622021993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,3584,0.0794272025426229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,4096,0.03961706558863322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,3584,0.03465813398361206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,3072,0.017778132359186807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,3072,0.06496320168177286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,768,65536,0.4922282536824544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,2560,0.015770666797955833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,2560,0.05383786757787069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,3072,0.030096000432968138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,2048,0.012846933801968894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,2048,0.04367253383000692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,2560,0.02648106614748637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,1536,0.01050879955291748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,1536,0.037212800979614255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,2048,0.022261333465576173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,1024,0.008553600311279297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,1024,0.03157013257344564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,768,0.0067552000284194945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,768,0.02717333237330119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,1536,0.01876373291015625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,512,0.005914666752020518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,512,0.02387626568476359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,768,0.012806399663289388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,1024,0.015050666530927024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,256,0.004667733112970988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,256,0.023126399517059325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,512,0.013377066453297934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,128,0.004334933559099833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,256,0.016115199526151022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,64,0.003870933254559835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,128,0.0191210667292277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,512,32,0.0042015999555587765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,128,0.015869866808255514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,64,0.018551466862360637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,512,32,0.018937599658966065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,65536,0.20343359311421713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,16384,0.057879467805226646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,16384,0.3940160115559896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,65536,1.5194890340169271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,12288,0.041970133781433105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,12288,0.29815572102864585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,16384,0.10537493228912354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,10240,0.03711893161137898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,12288,0.08397440115610758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,10240,0.2490613301595052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,8192,0.03097493251164754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,8192,0.2018634637196859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,10240,0.07585173447926839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,7168,0.025974400838216144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,7168,0.17513599395751953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,8192,0.06252479950586955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,6144,0.023135999838511147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,6144,0.15070826212565105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,7168,0.05852799812952677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,512,65536,0.3792959849039713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,5120,0.016541866461435954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,5120,0.12180159886678059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,6144,0.05529919862747192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,4096,0.013452800114949546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,4096,0.08830826282501221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,3584,0.01204373339811961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,3584,0.07242453098297119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,5120,0.05110506614049276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,3072,0.010342400272687275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,3072,0.05785706837972006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,4096,0.03840959866841634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,2560,0.00922986666361491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,2560,0.04734933376312256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,3584,0.03388906717300415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,2048,0.007889066636562348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,2048,0.04067946672439575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,3072,0.029741867383321123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,1536,0.006696533163388569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,2560,0.02647999922434489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,1536,0.03469226757685344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,1024,0.0053493330876032506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,1024,0.028827732801437377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,2048,0.02193066676457723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,768,0.004692266881465912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,1536,0.018655999501546224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,768,0.02688639958699544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,512,0.004346666733423868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,1024,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,512,0.022722133000691733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,768,0.01291306714216868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,256,0.0035391998787721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,512,0.01320319970448812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,256,0.020601600408554077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,128,0.0030250666042168934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,256,0.015822933117548624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,128,0.018595200777053834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,64,0.0030975999931494398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,128,0.015745066603024802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,256,32,0.003089066594839096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,64,0.017040000359217326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,256,32,0.01752106746037801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,65536,0.17051199277242024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,16384,0.047622398535410566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,16384,0.38458665211995446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,65536,1.5135626475016275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,256,65536,0.3677066802978516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,12288,0.03431893189748128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,12288,0.2918485323588053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,16384,0.11585386594136556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,10240,0.029614933331807453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,10240,0.2439722696940104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,12288,0.09104106426239014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,8192,0.024530132611592613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,8192,0.1971903959910075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,10240,0.08279146353403727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,7168,0.021192532777786256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,7168,0.1717631975809733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,8192,0.06685012976328532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,6144,0.01586026648680369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,6144,0.14752000172932941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,5120,0.011708799997965496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,7168,0.0630517323811849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,5120,0.11937173207600911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,6144,0.05796800057093302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,4096,0.009410132964452107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,4096,0.086790402730306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,3584,0.008412800232569377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,5120,0.05383573373158773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,3584,0.07073173522949219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,3072,0.007482666770617168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,4096,0.03996479908625285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,3072,0.05548799832661947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,2560,0.006687999765078227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,65536,0.3907125473022461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,3584,0.036185598373413085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,2048,0.005570133527119955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,2560,0.04747519890467326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,2048,0.039094400405883786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,3072,0.03171413342158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,1536,0.005009066561857859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,2560,0.027643734216690065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,1536,0.033297065893809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,2048,0.023151999711990355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,1024,0.004251733422279358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,768,0.00377813329299291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,1024,0.02722559968630473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,1536,0.01941653291384379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,512,0.0033802665770053864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,768,0.025144533316294355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,1024,0.014893866578737893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,256,0.003035733352104823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,768,0.013237333297729493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,512,0.02185386617978414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,128,0.0026367999613285064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,256,0.020524799823760986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,512,0.01322879989941915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,64,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,256,0.016105600198109946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,128,0.01729493339856466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,128,32,0.0027701333165168762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8192,128,128,0.015675733486811318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,64,0.017388800779978432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,128,32,0.01518186628818512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,16384,0.04680533409118652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,65536,0.16256213188171387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,12288,0.03282133340835571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,10240,0.02804693380991618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,16384,0.3837770779927572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,12288,0.29079360961914064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,8192,0.023350399732589722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,10240,0.24385706583658853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,7168,0.020222934087117513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,6144,0.014869333306948344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,8192,0.19619733492533367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,7168,0.17124586105346679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,5120,0.009684266646703084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,6144,0.14627200762430828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,4096,0.007909333209196727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,3584,0.007145600020885467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,5120,0.11749440034230549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,4096,0.08622612953186035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,3072,0.006275199850400289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,3584,0.06905919710795085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,2560,0.006253866851329804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,2048,0.00544213354587555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,3072,0.05434240102767944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,2560,0.04524053335189819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,1536,0.00468800018231074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,1024,0.003815466662247976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,2048,0.03905386527379354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,768,0.003697066754102707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,1536,0.03358293374379476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,1024,0.027024000883102417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,512,0.0030559999247392017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,256,0.0026389333109060925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,768,0.025618133942286174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,65536,1.505254364013672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,128,0.002584533393383026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,512,0.022217599550882976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,64,0.0022133332987626392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,256,0.020261333386103312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,64,32,0.002583466718594233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,64,0.01636799971262614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,128,0.019000534216562906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,64,32,0.015495466192563376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,16384,0.04997760057449341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,65536,0.16107199986775717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,12288,0.031175466378529866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,16384,0.3840010643005371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,10240,0.026800000667572023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,12288,0.29119466145833334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,8192,0.022184532880783082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,10240,0.24439892768859864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,8192,0.19573866526285807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,7168,0.019435733556747437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,6144,0.01316373348236084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,7168,0.17204480171203612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,5120,0.008760533730189006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,6144,0.14620906511942547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,4096,0.007410133381684621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,5120,0.11794453461964924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,4096,0.08553067048390707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,3584,0.00673280010620753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,3072,0.0059232001503308615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,65536,1.491919962565104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,3584,0.06931520303090413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,2560,0.005419733126958212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,3072,0.05348586638768514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,2048,0.004637866715590159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,1536,0.004640000065167745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,2560,0.045076266924540205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,2048,0.03913280169169108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,1024,0.003852800031503042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,768,0.003455999990304311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,1536,0.03287893335024516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,1024,0.028438399235407513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,512,0.0030048000315825146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,256,0.002631466587384542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,768,0.025719465812047322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,128,0.002573866645495097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,512,0.02287893295288086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,256,0.01930026610692342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,64,0.0022463999688625337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,128,0.017753599087397258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8192,32,32,0.002253866692384084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,64,0.016911999384562174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8192,32,32,0.015877333283424378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,16384,4.500805155436198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,12288,6.625129699707031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,16384,8.195710754394531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,12288,3.4685674031575524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,10240,3.0371307373046874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,10240,6.053223673502604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,8192,4.597198994954427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,8192,2.3318837483723955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,8192,0.8688746770222983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,7168,3.863371785481771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,7168,1.8496159871419269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,10240,1.3476800282796224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,6144,3.289794158935547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,6144,1.593378194173177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,12288,1.729762140909831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,5120,2.884651692708333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,5120,1.1876256306966146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,6144,0.7340063730875651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,4096,2.1416234334309894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,4096,0.6818901062011719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,7168,0.8404256184895834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,3584,1.664251708984375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,3584,0.8249589284261069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,5120,0.5796192169189454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,3072,1.6723808288574218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,3072,0.7524864196777343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,16384,2.46223882039388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,2560,1.1223499298095703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,2560,0.5791680018107097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,3584,0.42797333399454757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,2048,0.5756373087565104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,2048,0.3745866775512695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,4096,0.47896747589111327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,1536,0.5966698964436848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,1536,0.36082347234090173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,3072,0.3772981325785319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,1024,0.29268051783243815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,1024,0.20634986559549967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,2560,0.33063678741455077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,768,0.23623785972595215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,768,0.17120639483133954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,1024,0.1775551954905192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,512,0.17413652737935384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,512,0.14176106452941895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,768,0.15359360376993816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,256,0.11923519770304362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,256,0.11497920354207355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,256,0.10594027042388916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,128,0.10110399723052979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,128,0.10142186482747395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,512,0.13443093299865722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,64,0.08410987059275309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,64,0.10115093390146893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,65536,32,0.08611199855804444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,65536,32,0.10132053693135579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,1536,0.23982292811075845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,128,0.09320639769236247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,65536,5.076841735839844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,65536,9.152816772460938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,16384,1.8989461263020833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,16384,1.0137439727783204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,12288,0.8710282643636068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,12288,0.7418570836385091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,65536,2048,0.2827701250712077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,10240,0.8451530456542968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,10240,0.754745610555013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,10240,0.30532054901123046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,8192,0.6168928146362305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,8192,0.4610293388366699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,12288,0.35455681482950846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,7168,0.5187946637471517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,7168,0.5677226384480794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,16384,0.46146666208903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,6144,0.5197930653889974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,6144,0.4255861282348633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,8192,0.24874560038248697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,5120,0.3702058792114258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,5120,0.3152362823486328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,7168,0.22052693367004395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,4096,0.2919989267985026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,4096,0.23215146064758302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,6144,0.19479680061340332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,3584,0.2550432046254476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,3584,0.2010495980580648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,4096,0.138590939839681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,3072,0.2205205281575521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,3072,0.15273386637369793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,5120,0.17042239507039386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,2560,0.18075839678446454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,2560,0.1349514643351237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,3584,0.12446080048878987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,2048,0.14470507303873698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,2048,0.10904746850331623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,2560,0.09708053270975749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,1536,0.11305279731750488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,1536,0.09122560024261475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,3072,0.11048533121744793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,1024,0.07893973191579183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,1024,0.06800533135732015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,1024,0.05347626606623331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,768,0.06269546747207641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,768,0.06314346790313721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,1536,0.06995200316111247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,512,0.04772053162256877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,512,0.053010133902231846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,768,0.046573865413665774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,256,0.03235519925753276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,256,0.0443125327428182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,2048,0.08279146353403727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,128,0.028328533967336016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,128,0.03800959984461467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,256,0.0346346656481425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,64,0.023333332935969033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,64,0.0377461314201355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,16384,32,0.024154667059580484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,16384,32,0.03760000069936116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,128,0.03176533381144206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,512,0.04215253194173177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,65536,3.596245320638021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,16384,1.5860543568929037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,65536,6.265982055664063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,16384,0.8145578384399415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,12288,0.9597087860107422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,12288,0.6709023793538411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,12288,0.2877066612243652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,10240,0.7868778864542644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,10240,0.6256341298421224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,16384,0.37483094533284506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,8192,0.4411946614583333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,8192,0.5215903917948406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,10240,0.2456597328186035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,7168,0.37981014251708983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,7168,0.4250048001607259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,8192,0.2000981330871582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,6144,0.327344004313151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,6144,0.27827094395955404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,7168,0.18029653231302897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,5120,0.2695647875467936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,5120,0.220851198832194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,6144,0.15907840728759765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,4096,0.21846186319986977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,4096,0.16102080345153807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,5120,0.1385248025258382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,3584,0.18930773735046386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,3584,0.1471157391866048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,3584,0.100819198290507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,3072,0.1629205385843913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,3072,0.12638506889343262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,4096,0.11257706483205158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,2560,0.13924266497294108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,2560,0.11028587023417155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,2560,0.07887893517812093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,2048,0.11208319664001465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,2048,0.0896128018697103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,3072,0.09045226573944092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,1536,0.0852725346883138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,1536,0.07362133661905924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,1536,0.056683735052744544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,1024,0.05924266576766968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,1024,0.0564192016919454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,2048,0.06718400319417318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,768,0.04775360027949015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,768,0.05236479838689169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,1024,0.04338239828745524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,512,0.03483413457870484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,512,0.051378134886423746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,768,0.03865600029627482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,256,0.030825599034627275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,256,0.04349973201751709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,256,0.02900586724281311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,512,0.0343018651008606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,128,0.01978666583697001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,128,0.03262719909350077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,64,0.017777067422866822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,64,0.03275093237559001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,12288,32,0.018337066968282065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,12288,32,0.031031467517217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,128,0.026294400294621784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,65536,2.7298314412434896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,65536,5.492805480957031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,16384,1.0402442932128906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,16384,0.7025898615519206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,16384,0.312006409962972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,12288,0.5574143727620442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,12288,0.5086890538533528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,16384,65536,2.9169995625813803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,10240,0.6400864283243816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,10240,0.5221397399902343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,12288,0.24050559997558593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,8192,0.3621962547302246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,8192,0.4627519925435384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,10240,0.2082304000854492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,7168,0.32197440465291344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,7168,0.2772992134094238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,8192,0.16924479802449544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,6144,0.2693002700805664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,6144,0.20519572893778482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,12288,65536,1.6616565704345703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,5120,0.3210058530171712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,5120,0.2285674730936686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,6144,0.13445547421773274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,4096,0.1789120038350423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,4096,0.1456778685251872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,7168,0.15129920641581218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,3584,0.16053120295206708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,3584,0.1301472028096517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,5120,0.11859412988026936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,3072,0.13660693168640137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,3072,0.11101439793904622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,3584,0.08697280089060465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,2560,0.11621867020924885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,2560,0.0991978645324707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,4096,0.09581653277079263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,2048,0.09199786980946859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,2048,0.07990293502807617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,2048,0.057501868406931556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,1536,0.0708031972249349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,1536,0.06596479813257852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,2560,0.0669322649637858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,1024,0.05102399984995524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,1024,0.051269332567850746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,3072,0.07711040178934733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,768,0.03979093233744303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,768,0.04760533173878988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,768,0.03361279964447021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,512,0.030424533287684123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,512,0.041544532775878905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,1024,0.037773867448170975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,256,0.021704532702763877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,256,0.03366506497065226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,1536,0.048757334550221756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,128,0.016740266482035318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,128,0.030321067571640013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,256,0.025126399596532185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,64,0.015466666221618653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,128,0.023572266101837158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,64,0.03035413424173991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,10240,32,0.016422399878501893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,10240,32,0.030039467414220172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,512,0.029872000217437744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,65536,2.487755839029948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,16384,0.8407797495524088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,65536,4.424406433105469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,16384,0.5474506378173828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,12288,0.6461845397949219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,12288,0.4518623987833659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,12288,0.20355733235677084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,10240,0.39919999440511067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,10240,0.38742186228434244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,16384,0.2652693271636963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,8192,0.3183690706888834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,8192,0.2858218510945638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,10240,0.17315732638041179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,7168,0.2556938648223877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,7168,0.2718816121419271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,8192,0.13829973538716633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,6144,0.21362560590108237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,6144,0.20071253776550294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,7168,0.12587839762369793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,5120,0.1813418706258138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,5120,0.15030399958292645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,6144,0.11064426898956299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,4096,0.14235946337382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,4096,0.12242666880289714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,10240,65536,1.5169717152913411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,3584,0.19043946266174316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,3584,0.11835839748382568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,5120,0.09882026513417562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,3072,0.11111040115356445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,3072,0.09776533444722493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,3584,0.07312426567077637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,2560,0.09288960297902425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,2560,0.08596479892730713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,4096,0.08055573304494222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,2048,0.07467412948608398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,2048,0.06974506378173828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,2560,0.05679466724395752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,3072,0.06577173471450806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,1536,0.05814506610234579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,1536,0.059531732400258386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,1024,0.04115093151728312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,2048,0.047777068614959714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,1024,0.044248533248901364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,768,0.03186986645062764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,768,0.041681067148844404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,768,0.028280532360076903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,512,0.023864533503850302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,512,0.03597866694132487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,1024,0.031114667654037476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,256,0.01660480002562205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,256,0.029763199885686237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,256,0.02145599921544393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,128,0.01548693378766378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,128,0.03196799953778585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,512,0.02528746724128723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,64,0.016989866892496742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,128,0.020403200387954713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,64,0.03102933367093404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,8192,32,0.017593600352605186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,8192,32,0.03012479941050212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,1536,0.039562666416168214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,65536,1.9391146341959637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,16384,0.7459530512491862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,65536,3.648443857828776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,16384,0.5444405237833659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,12288,0.4564522743225098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,12288,0.5073482513427734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,12288,0.17798080444335937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,10240,0.3173503875732422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,10240,0.2778208096822103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,16384,0.2324735959370931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,8192,0.2495434602101644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,8192,0.3675519943237305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,10240,0.15275519688924152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,7168,0.22261974016825356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,7168,0.21549013455708824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,8192,0.12171306610107421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,6144,0.1868842601776123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,6144,0.17092587153116862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,7168,0.11146879990895589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,5120,0.15800639788309734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,5120,0.138046932220459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,6144,0.09891626834869385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,4096,0.12410666942596435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,4096,0.1117674668629964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,5120,0.08807466824849447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,3584,0.11186880270640057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,3584,0.10456533432006836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,4096,0.07176746527353922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,3072,0.09432213306427002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,3072,0.08800960381825765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,3584,0.06548800071080527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,2560,0.08161493142445883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,2560,0.07914026578267416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,3072,0.05778133471806844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,2048,0.06504533290863038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,2048,0.06379626592000326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,2560,0.05020053386688232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,1536,0.051303466161092125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,1536,0.05372159878412882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,2048,0.04237866799036662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,1024,0.03447253306706746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,1024,0.040957868099212646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,1024,0.028104533751805622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,768,0.02680533329645793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,768,0.037581864992777506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,1536,0.03590933481852214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,512,0.0206496000289917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,512,0.032542934020360306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,768,0.02479040026664734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,256,0.014677332838376364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,256,0.026844799518585205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,512,0.022948267062505086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,128,0.013436800241470337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,256,0.019767467180887857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,128,0.02487893303235372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,64,0.011387733618418376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,64,0.02483200033505758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,7168,32,0.011700266599655151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,128,0.018198400735855103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,7168,32,0.024769065777460735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,65536,3.0405568440755206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,65536,1.7803690592447918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,16384,0.5415487925211588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,16384,0.5398687998453776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,8192,65536,1.0116714477539062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,12288,0.5186815897623698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,12288,0.424020258585612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,12288,0.1765877405802409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,10240,0.28889385859171546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,10240,0.2654154618581136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,16384,0.23077333768208824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,8192,0.23185386657714843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,8192,0.20586026509602867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,10240,0.15184000333150227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,7168,0.1997183958689372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,7168,0.17838293711344402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,8192,0.12202560106913249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,6144,0.17150400479634603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,6144,0.15333226521809895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,7168,0.11021119753519695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,5120,0.14167787233988444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,5120,0.1301695982615153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,6144,0.09834240277608236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,4096,0.11494399706522625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,4096,0.10666133562723797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,5120,0.08644159634908041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,3584,0.10004266897837322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,3584,0.09812800089518228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,4096,0.0707157293955485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,3072,0.08725972970326742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,3072,0.08335893154144287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,3584,0.06390933195749918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,2560,0.07249279816945395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,2560,0.07470613320668539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,3072,0.056074666976928714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,2048,0.05991679827372233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,2048,0.059432534376780186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,2560,0.048706134160359696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,1536,0.04559573332468669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,1536,0.04957866668701172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,2048,0.04107733170191447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,1024,0.03158506751060486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,1024,0.037598931789398195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,7168,65536,0.8942922592163086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,768,0.024676267306009927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,768,0.03712853193283081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,1024,0.02714879910151164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,512,0.018755199511845906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,512,0.030954666932423908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,1536,0.034306132793426515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,256,0.01360106666882833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,768,0.024770132700602212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,256,0.026808534065882367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,128,0.012039466698964437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,128,0.025893332560857137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,128,0.018173867464065553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,64,0.010012800494829815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,64,0.024142932891845704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,6144,32,0.01076800028483073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,256,0.019027199347813925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,6144,32,0.02285226583480835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,512,0.022247467438379923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,65536,1.5112309773763022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,65536,2.525429280598958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,16384,0.38739093144734704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,16384,0.5526218414306641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,12288,0.29737812678019204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,12288,0.36050132115681965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,12288,0.15019413630167644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,10240,0.24349652926127113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,10240,0.24578879674275717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,16384,0.19820906321207682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,8192,0.1994592030843099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,8192,0.19346240361531575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,10240,0.13029440244038898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,7168,0.17201919555664064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,7168,0.16701547304789227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,8192,0.1066431999206543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,6144,0.1496064027150472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,6144,0.14317973454793292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,7168,0.09561813672383626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,5120,0.12270186742146809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,5120,0.12045760154724121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,6144,0.08646079699198404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,4096,0.09940799872080484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,4096,0.09994346300760905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,5120,0.07674880027770996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,3584,0.08685759703318277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,3584,0.09006079832712809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,4096,0.0615231990814209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,3072,0.07539947032928467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,3072,0.07701546351114909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,3584,0.05574719905853272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,2560,0.06301973263422647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,2560,0.0670464038848877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,3072,0.049097601572672525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,2048,0.051507198810577394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,2048,0.05443946520487467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,2560,0.04267839988072713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,1536,0.04001706838607788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,1536,0.0474239985148112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,2048,0.03594026565551758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,1024,0.027261867125829058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,1024,0.03776106834411621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,1536,0.030469334125518797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,768,0.02185813387235006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,768,0.03558506568272908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,1024,0.023922133445739745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,512,0.018317866325378417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,512,0.032631466786066696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,768,0.021923200289408366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,256,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,256,0.02501973311106364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,256,0.017348267634709678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,128,0.009153067072232565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,512,0.02021333376566569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,128,0.02173759937286377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,64,0.008312533299128216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,64,0.023042132457097374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,5120,32,0.01037013332049052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,5120,32,0.02277759909629822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,128,0.016581333676973977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,65536,1.3577173868815104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,65536,1.973849614461263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,6144,65536,1.0271029154459634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,16384,0.4837183952331543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,16384,0.4370314598083496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,12288,0.2666261355082194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,12288,0.2860447883605957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,12288,0.12716586589813234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,10240,0.19717119534810384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,10240,0.21945813496907554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,16384,0.16376852989196777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,8192,0.15856107076009113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,8192,0.16683732668558757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,10240,0.11077653566996257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,7168,0.13926186561584472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,7168,0.14356692632039386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,8192,0.09019200007120767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,6144,0.11979520320892334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,6144,0.12300159931182861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,7168,0.0815071980158488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,5120,0.0956063985824585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,5120,0.10437333583831787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,6144,0.07344213326772055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,4096,0.07834133307139078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,4096,0.08563946882883708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,5120,0.0644597331682841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,3584,0.06801066398620606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,3584,0.07857813040415446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,4096,0.05256213347117106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,3072,0.05937173366546631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,3072,0.06604693333307901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,3584,0.04708053270975749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,2560,0.04951359828313191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,2560,0.05753599802652994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,2560,0.036448001861572266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,2048,0.04017173449198405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,2048,0.04782720009485881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,3072,0.04275946617126465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,1536,0.03176533381144206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,1536,0.041427199045817056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,1536,0.026114133993784587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,1024,0.02218559980392456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,1024,0.03316799998283386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,2048,0.03065920074780782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,768,0.017284266153971353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,768,0.029950932661692305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,768,0.019282132387161255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,512,0.012753066420555115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,512,0.026962133248647054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,1024,0.020811732610066733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,256,0.009614933530489604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,256,0.023845332860946655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,512,0.017722666263580322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,128,0.009545600414276123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,256,0.01585493286450704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,128,0.02086506684621175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,64,0.007537066439787547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,64,0.020592000087102255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,4096,32,0.007760000228881836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,4096,32,0.020920532941818237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,128,0.015347199638684592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,65536,1.6239935557047527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,65536,1.3199317932128907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,16384,0.28724587758382164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,16384,0.34465599060058594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,5120,65536,0.7411231994628906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,12288,0.2449941317240397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,12288,0.2682058652242025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,12288,0.10513813495635986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,10240,0.16037972768147785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,10240,0.19812480608622235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,16384,0.13538133303324382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,8192,0.12753067016601563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,8192,0.15048747062683104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,10240,0.09229013125101725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,7168,0.11324053605397541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,7168,0.13274772961934406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,8192,0.07447146574656169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,6144,0.09741973082224528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,6144,0.11325653394063313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,6144,0.06171093384424845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,5120,0.08089813391367594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,7168,0.069541335105896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,5120,0.09545706907908122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,4096,0.06491733392079671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,4096,0.07819519837697347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,4096,0.04425599972407023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,3584,0.056873599688212075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,3584,0.07003520329793295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,5120,0.054024533430735266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,3072,0.04866559902826945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,3072,0.059324800968170166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,3584,0.04006933371225993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,2560,0.041316266854604086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,3072,0.03575253486633301
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,2560,0.05132053295771281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,2048,0.03369706471761068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,2048,0.04323519865671794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,4096,65536,0.6152373631795247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,1536,0.026038400332132977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,1536,0.03747626543045044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,2048,0.026415999730428057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,1024,0.018849066893259683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,1024,0.03081066608428955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,2560,0.03095146616299947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,768,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,768,0.02879679997762044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,768,0.016921599706014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,512,0.010708266496658325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,512,0.026037333408991496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,1024,0.017618133624394735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,256,0.008111999928951263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,256,0.022791467110315957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,1536,0.022316799561182658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,128,0.007125333448251088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,128,0.021769599119822184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,256,0.014502400159835815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,64,0.006739200154940288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,64,0.020552533864974975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3584,32,0.007077333331108093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,512,0.016100266575813295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3584,32,0.020565332969029744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,128,0.01565439999103546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,65536,1.0870442708333334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,65536,1.2142101287841798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,16384,0.3900906562805176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,16384,0.316103458404541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,12288,0.17743786176045734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,12288,0.23875734011332192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,12288,0.10424319903055827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,10240,0.1513482729593913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,10240,0.18973654111226398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,16384,0.13402026494344074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,8192,0.12021760145823161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,8192,0.14559787114461262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,10240,0.09145600001017253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,7168,0.10690240065256755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,7168,0.12700053056081134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,8192,0.07366399765014649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,6144,0.09023359616597494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,6144,0.1076586643854777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,7168,0.06856640179951987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,5120,0.07362346649169922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,5120,0.0902026653289795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,6144,0.06012586752573649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,4096,0.05868586699167887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,4096,0.07392106850941976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,5120,0.05264533360799154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,3584,0.05207573175430298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,3584,0.0655893325805664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,4096,0.04297600189844768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,3072,0.044728533426920576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,3072,0.05626240173975626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,3584,0.03924479881922404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,2560,0.0380789319674174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,2560,0.04979519844055176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,3072,0.035086933771769205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,2048,0.030740267038345336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,2048,0.04204053481419881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,2560,0.030624000231424968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,1536,0.024158932765324912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,1536,0.0368341326713562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3584,65536,0.500712521870931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,1024,0.016985599199930826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,1024,0.03075946569442749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,1536,0.022318933407465616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,2048,0.02592746615409851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,768,0.013974400361378989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,768,0.028899200757344562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,512,0.011570133765538533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,512,0.02685653368631999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,512,0.01581333378950755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,768,0.016635732849438985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,256,0.009284266829490661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,1024,0.01771519978841146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,256,0.024834134181340537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,128,0.009064533313115438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,128,0.0228383998076121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,256,0.014594133694966635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,64,0.00842133363087972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,128,0.014079999923706055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,3072,32,0.008769067128499348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,64,0.02174826661745707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,3072,32,0.02099306583404541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,65536,0.8089216232299805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,65536,1.1405323028564454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,16384,0.20385279655456542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,16384,0.2983349482218424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,12288,0.15310932795206705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,12288,0.22466880480448403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,12288,0.10438293615976971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,10240,0.12401173114776612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,10240,0.1776682694753011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,16384,0.13446399370829265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,8192,0.10128426551818848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,8192,0.136353063583374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,10240,0.09091306527455648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,7168,0.08546880086263021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,7168,0.11621226469675701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,8192,0.07358506520589193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,6144,0.07432320117950439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,6144,0.09974719683329264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,7168,0.06742719809214273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,5120,0.06097919940948486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,5120,0.08385279973347982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,6144,0.05999253193537394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,4096,0.05019840002059937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,4096,0.06634666522343954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,5120,0.05177599986394247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,3584,0.04344000021616618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,3584,0.05981333255767822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,3584,0.038734932740529374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,3072,0.03747946818669637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,3072,0.05130453507105509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,4096,0.042637864748636886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,2560,0.03250773350397746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,2560,0.0463861346244812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,2560,0.030778666337331135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,2048,0.026636799176534016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,2048,0.03940373261769613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,3072,0.034970664978027345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,1536,0.020061866442362467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,1536,0.03532586495081584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,1536,0.021891200542449953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,1024,0.014223999778429666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,1024,0.028875732421875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,2048,0.026077866554260254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,768,0.012344533205032348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,768,0.027639466524124145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,1024,0.01731946667035421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,512,0.010233599940935771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,512,0.025525333484013875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,768,0.01652479966481527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,256,0.008331733445326488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,3072,65536,0.49100799560546876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,256,0.02392639915148417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,512,0.015319466590881348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,128,0.007575466732184092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,128,0.020932267109553017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,64,0.006706133484840393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,64,0.01988053321838379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2560,32,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,128,0.013682132959365845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,256,0.01416853368282318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2560,32,0.02005973259607951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,65536,0.5975050608317057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,16384,0.24807467460632324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,65536,1.038101323445638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,16384,0.3251530647277832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,12288,0.12318186759948731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,12288,0.21189653078715004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,12288,0.08039893309275309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,10240,0.10195733706156414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,10240,0.16831893920898439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,16384,0.10264639854431153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,8192,0.08134400049845378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,8192,0.12887787024180095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,10240,0.07148053646087646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,7168,0.07167999744415283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,7168,0.11017066637674969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,8192,0.05803413391113281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,6144,0.061015466849009194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,6144,0.09351893266042074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,7168,0.05318613449732462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,5120,0.05095146497090658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,5120,0.07712852954864502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,6144,0.04708480040232341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,4096,0.041187198956807454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,4096,0.06126506725947062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,5120,0.041468799114227295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,3584,0.03695253531138103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,3584,0.05511786540349325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,4096,0.03467626571655273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,3072,0.03169599970181783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,3072,0.048585601647694904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,3584,0.03134079972902934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,2560,0.02698880036671956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,2560,0.043988267580668136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,3072,0.028075732787450153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,2048,0.02241493264834086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,2048,0.037573333581288657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,2560,0.025149865945180254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,2048,0.021154133478800456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,1536,0.01790613333384196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,1536,0.03426986535390218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,1024,0.01279146671295166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,1024,0.028161066770553588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,1024,0.01456106702486674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,768,0.010841600100199382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,768,0.026850134134292603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,1536,0.018196266889572144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,512,0.00844586690266927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,512,0.02476480007171631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,512,0.013802666465441385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,768,0.014111999670664468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,256,0.0069578667481740315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,256,0.022487467527389525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,128,0.005937066674232483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,128,0.0211082657178243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2560,65536,0.4908298810323079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,64,0.005891199906667074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,256,0.012844799955685934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,2048,32,0.006301866471767425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,64,0.01869759956995646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,2048,32,0.01872426668802897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,128,0.012717866897583007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,65536,0.46124693552652996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,16384,0.18820266723632811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,65536,0.9663520177205405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,16384,0.2970634778340658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,12288,0.08951573371887207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,12288,0.18906453450520833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,12288,0.08040853341420492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,10240,0.08731413682301839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,10240,0.1560512065887451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,16384,0.10250133673350018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,8192,0.06994986534118652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,8192,0.11881386439005534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,10240,0.07148586908976237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,7168,0.060065066814422606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,7168,0.10086293220520019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,8192,0.057358932495117185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,6144,0.050904532273610435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,6144,0.08564480145772299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,7168,0.05179946819941202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,5120,0.041262932618459064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,5120,0.0707477331161499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,6144,0.0470090667406718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,4096,0.03300906618436177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,4096,0.05530666510264078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,5120,0.040907732645670575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,3584,0.028808534145355225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,3584,0.05118506749471029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,4096,0.03421760002772013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,3072,0.025542400280634564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,3072,0.044411734739939375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,3584,0.031274666388829545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,2560,0.021361066897710165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,2560,0.04034133354822795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,3072,0.027406932910283406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,2048,0.017630932728449504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,2048,0.03502080043156942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,2560,0.024361600478490196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,1536,0.013768532872200012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,1536,0.030935466289520264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,1536,0.01819093426068624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,2048,0.021077332894007365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,1024,0.010412800312042236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,1024,0.02677653431892395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,768,0.008756267031033833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,768,0.025124265750249224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,768,0.014064000050226847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,512,0.007163733243942261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,1024,0.014137599865595499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,512,0.022680532932281495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,256,0.005799466868241628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,256,0.020679465929667153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,512,0.01365013321240743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,256,0.012836266557375589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,128,0.0050687998533248905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,128,0.020129066705703736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,64,0.00505920002857844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,128,0.012414933244387309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,64,0.01863893270492554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1536,32,0.005037866532802582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1536,32,0.018218666315078735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,65536,0.3136106808980306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,65536,0.8803328196207681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,16384,0.08125226497650147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,16384,0.22869440714518227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,2048,65536,0.372602653503418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,12288,0.06121493180592855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,12288,0.1704352060953776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,12288,0.057494401931762695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,10240,0.05135253270467123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,10240,0.1397088050842285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,16384,0.07296746571858724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,8192,0.0440778652826945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,8192,0.10699306329091389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,10240,0.05182826519012451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,7168,0.03705386718114217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,7168,0.08873813152313233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,8192,0.042532265186309814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,6144,0.03254826664924622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,6144,0.07508800029754639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,7168,0.03823146820068359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,5120,0.027036799987157183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,5120,0.05966080029805502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,6144,0.036060798168182376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,5120,0.030589866638183593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,4096,0.02251840035120646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,4096,0.04845759868621826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,3584,0.019671465953191122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,3584,0.045193600654602054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,3584,0.02405440012613932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,3072,0.017423999309539796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,3072,0.0410805344581604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,4096,0.026210133234659833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,2560,0.015161599715550741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,2560,0.03816106716791789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,3072,0.02109760046005249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,2048,0.012891733646392822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,2048,0.032013867298762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,2560,0.019057067235310872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,1536,0.010498133301734925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,1536,0.02890239953994751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,1536,0.01422826647758484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,2048,0.01651946703592936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,1024,0.008154666423797608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,1024,0.024152533213297526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,768,0.006762666503588359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,768,0.02459519902865092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,1024,0.012110933661460876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,768,0.012011733651161195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,512,0.005784533421198527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,512,0.02090239922205607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,256,0.004648533463478088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,256,0.019106133778889974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,512,0.012181333700815837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,256,0.011588266491889954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,128,0.0038762666285037995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,128,0.01762346625328064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,64,0.003833599885304769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,64,0.01863893270492554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,1024,32,0.0038560000558694207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,128,0.011353600025177001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,1024,32,0.01853653391202291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,65536,0.24518613815307616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,65536,0.8501525243123373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,16384,0.06297599871953329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,16384,0.2185376008351644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1536,65536,0.3699743906656901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,12288,0.04953386783599854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,12288,0.16440107027689616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,12288,0.057523198922475184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,10240,0.04163839817047119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,16384,0.07225813070933024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,10240,0.1349290688832601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,8192,0.03377813498179118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,8192,0.10269227027893066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,8192,0.040972801049550374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,7168,0.0294815997282664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,10240,0.0516490658124288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,7168,0.08566933472951253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,6144,0.025638399521509807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,6144,0.07001813252766928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,7168,0.038100266456604005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,5120,0.02122453252474467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,5120,0.05529280106226603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,6144,0.033829331398010254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,4096,0.017847466468811034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,4096,0.04728000164031983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,1024,65536,0.2542474587758382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,3584,0.016588800152142844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,3584,0.04297706683476766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,5120,0.030885332822799684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,3072,0.013756799697875976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,3072,0.03773760000864665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,3584,0.023484800259272257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,4096,0.02592746615409851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,2560,0.011889066298802693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,2560,0.03601919809977214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,2048,0.010452266534169514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,2048,0.030986666679382324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,2560,0.018978132804234823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,1536,0.00851200024286906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,3072,0.02142613331476847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,2048,0.016411733627319337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,1536,0.02880000074704488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,1024,0.0067669332027435304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,1024,0.02434879938761393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,768,0.00589333325624466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,768,0.02336853345235189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,1024,0.012071466445922852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,1536,0.01414293348789215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,512,0.005048533280690512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,512,0.02090453306833903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,768,0.0116757333278656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,256,0.0038890667259693147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,256,0.018668800592422485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,512,0.012151466806729634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,128,0.0038101332883040107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,128,0.016902399063110352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,256,0.011705600221951802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,64,0.0034111998975276947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,128,0.011215999722480774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,64,0.017362133661905924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,768,32,0.004660266637802124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,768,32,0.016540799538294473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,65536,0.17880746523539226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,65536,0.8141301472981771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,16384,0.04865066607793172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,16384,0.21106452941894532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,12288,0.0375925342241923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,12288,0.158243195215861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,12288,0.0572213331858317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,10240,0.030585600932439165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,10240,0.1287402629852295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,16384,0.07158079942067465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,8192,0.02414720058441162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,8192,0.09532693227132162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,10240,0.05142506758371988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,7168,0.020406399170557657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,7168,0.07789759635925293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,8192,0.04127999941507975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,6144,0.01817493240038554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,6144,0.06365973154703776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,7168,0.03689493338267009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,5120,0.015201066931088766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,6144,0.033353598912556966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,5120,0.0513322671254476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,4096,0.012568533420562744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,4096,0.04310400088628133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,5120,0.030921600262324017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,4096,0.025540266434351606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,3584,0.01162559986114502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,3584,0.040814932187398276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,3072,0.010021332899729412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,3072,0.03646293481190999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,3584,0.023549866676330567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,3072,0.02104640007019043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,2560,0.009199999769528707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,2560,0.03314346671104431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,2048,0.007626666625340779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,2048,0.02905706763267517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,768,65536,0.25026987393697103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,1536,0.006650666892528534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,1536,0.027009065945943194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,2560,0.018567466735839845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,1024,0.005133866767088572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,2048,0.016102400422096253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,1024,0.023171200354894003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,768,0.00470719983180364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,768,0.02268480062484741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,1024,0.012005333105723064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,512,0.0038389332592487337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,1536,0.014552533626556396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,512,0.020658133427302043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,256,0.003385599950949351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,768,0.011646933356920878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,256,0.01920106609662374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,512,0.012008532881736755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,128,0.0030591999491055804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,256,0.011687466502189636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,128,0.01660053332646688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,64,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,128,0.010818133751551311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,512,32,0.0030303999781608583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,64,0.015864533185958863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,512,32,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,65536,0.11020479996999104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,16384,0.037460267543792725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,16384,0.20398186047871908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,65536,0.7734975814819336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,12288,0.028173865874608357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,12288,0.15383893648783367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,12288,0.05748053391774496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,10240,0.021228800217310585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,16384,0.07103679974873861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,10240,0.12325332959493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,8192,0.017909334103266398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,8192,0.09127360184987386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,10240,0.05117119948069254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,7168,0.015153066317240397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,8192,0.04078720013300578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,7168,0.07359146277109782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,6144,0.013185066978136697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,6144,0.0585375984509786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,7168,0.03672639926274617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,5120,0.011384532848993937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,6144,0.03322239915529887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,5120,0.04783039887746175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,4096,0.009648000200589497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,4096,0.04120746850967407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,5120,0.030499200026194256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,3584,0.008699733018875121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,4096,0.025886933008829754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,3584,0.03902399937311808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,512,65536,0.2448575973510742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,3072,0.007973333199818928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,3072,0.03546026547749837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,3584,0.02359786629676819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,2560,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,2560,0.032919466495513916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,3072,0.021154133478800456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,2048,0.005973333120346069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,2048,0.028832000494003297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,2560,0.018982400496800743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,1536,0.005485866467158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,1536,0.026858667532602947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,2048,0.01614293356736501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,1024,0.004701866706212362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,1024,0.022707200050354003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,1536,0.014200533429781595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,768,0.003907199949026108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,768,0.02250880002975464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,1024,0.011658666531244914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,512,0.003505066782236099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,512,0.020435200134913126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,768,0.01200320025285085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,256,0.0030933332939942675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,512,0.011986133456230164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,256,0.01814080079396566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,128,0.002616533388694127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,128,0.016338133811950685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,64,0.002569599946339925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,256,0.01167466640472412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,128,0.011115733782450359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,256,32,0.0026933332284291584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,64,0.016327466567357382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,256,32,0.014677332838376364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,65536,0.09861653645833333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,16384,0.02886613408724467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,16384,0.2005077362060547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,65536,0.7625941594441732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,12288,0.0213919997215271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,256,65536,0.24453652699788414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,12288,0.14924906094868978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,10240,0.015175466736157736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,10240,0.12114986578623455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,12288,0.05637973149617513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,16384,0.07167786757151286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,8192,0.012117333213488261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,8192,0.08767573038736978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,10240,0.05025173425674438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,7168,0.01099626620610555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,7168,0.07081813017527262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,6144,0.009687466422716777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,8192,0.039476267496744794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,6144,0.05572906732559204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,7168,0.03658239841461182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,5120,0.00844693382581075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,5120,0.04690026839574178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,4096,0.0071946665644645694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,4096,0.03948479890823364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,6144,0.033319467306137086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,3584,0.0066890666882197065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,5120,0.03057173291842143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,3584,0.03756693204243978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,3072,0.006279466549555461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,3072,0.0337557315826416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,4096,0.025572266181310015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,2560,0.0056415999929110205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,3584,0.023129600286483764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,2560,0.03240426580111186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,2048,0.004666666686534882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,2048,0.02897599935531616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,3072,0.020874667167663574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,1536,0.004262400170167288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,2560,0.018601600329081217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,1536,0.02602880001068115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,1024,0.00342399999499321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,2048,0.016127999623616537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,1024,0.022447999318440756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,1536,0.014079999923706055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,768,0.0033610666791598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,1024,0.011602133512496948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,768,0.02143999934196472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,512,0.0030037333567937215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,768,0.011613866686820984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,512,0.02001813252766927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,256,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,512,0.01165226697921753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,256,0.017951999107996622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,128,0.002553600072860718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,256,0.011187199751536052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,128,0.016578132907549538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,64,0.002229333420594533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,65536,0.2448106606801351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4096,128,128,0.011224533120791119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,128,32,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,64,0.015156267086664834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,128,32,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,65536,0.09178453286488851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,16384,0.025616000096003216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,12288,0.017617066701253258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,10240,0.014075733224550881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,16384,0.19954880078633624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,12288,0.1478805383046468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,8192,0.011655466755231221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,10240,0.12075200080871581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,8192,0.08646293481191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,7168,0.010474666953086853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,6144,0.00920960009098053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,65536,0.7623168309529622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,7168,0.06967999935150146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,5120,0.007952000200748443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,6144,0.05709226528803507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,4096,0.006678399940331777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,5120,0.04543893337249756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,3584,0.006277333199977875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,3072,0.005868799984455109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,4096,0.04004160165786743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,3584,0.03704426685969035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,2560,0.005095466474692027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,3072,0.03355093399683635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,2048,0.0046304002404212955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,1536,0.003894400099913279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,2560,0.032279467582702635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,2048,0.027765333652496338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,1024,0.003504000107447306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,1536,0.025748266776402788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,768,0.003091199944416682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,1024,0.021335466702779134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,512,0.0029824001093705496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,256,0.0026752000053723653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,768,0.021601066986719767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,512,0.020041600863138834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,128,0.0022154666483402254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,256,0.01841493248939514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,64,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,128,0.0159850666920344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,64,32,0.002277333289384842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,64,0.014825600385665893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,64,32,0.014999467134475707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,65536,0.0871509313583374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,16384,0.025959465901056928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,12288,0.015451733271280924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,16384,0.19759893417358398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,12288,0.14857494036356608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,10240,0.011201066772143047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,8192,0.009336533149083455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,10240,0.1218453327814738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,7168,0.008734933535257975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,8192,0.08809599876403809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,65536,0.7545493443806965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,6144,0.007987200220425924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,7168,0.0722261349360148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,5120,0.011125333110491435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,4096,0.009586133559544881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,5120,0.04724160035451253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,6144,0.05615253448486328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,3584,0.008377599716186523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,3072,0.00767680009206136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,4096,0.039204267660776775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,3584,0.03893653154373169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,2560,0.006806399921576183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,2048,0.004724266628424326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,3072,0.03468480110168457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,1536,0.0038719999293486277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,2560,0.03238720099131266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,1024,0.0034005333979924522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,2048,0.027193599939346315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,1536,0.02575040062268575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,768,0.003014400104681651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,1024,0.022898133595784506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,512,0.002690133452415466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,256,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,512,0.019334399700164796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,768,0.021262933810551964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,128,0.0022730665902296705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,256,0.017373865842819212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,64,0.0022485333184401194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,128,0.016451199849446617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4096,32,32,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,64,0.015812266866366068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4096,32,32,0.014979199568430582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,16384,2.2553802490234376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,12288,3.1408353169759113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,16384,4.118246459960938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,12288,1.2006431579589845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,10240,1.0835402170817057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,10240,2.807049560546875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,8192,2.1479530334472656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,8192,0.9653258641560873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,8192,0.44190400441487626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,7168,1.8106048583984375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,7168,0.7346133550008138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,10240,0.5403647740681966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,6144,1.637882614135742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,6144,0.4781333287556966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,12288,0.6485375722249349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,5120,1.2375754038492839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,5120,0.46413227717081706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,6144,0.34386345545450847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,4096,0.5646293640136719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,4096,0.4440330823262532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,7168,0.39560960133870443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,3584,0.6211264292399089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,3584,0.375982920328776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,5120,0.3006186803181966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,3072,0.4205343882242839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,3072,0.3981397310892741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,16384,0.8388394673665365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,2560,0.46517759958903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,2560,0.2854698816935221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,3584,0.22220373153686523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,2048,0.2837279955546061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,2048,0.2748415946960449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,4096,0.24400639533996582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,1536,0.21438186963399253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,1536,0.15577066739400228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,3072,0.19694080352783203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,1024,0.1519584019978841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,1024,0.11116586526234944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,2560,0.16855252583821614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,768,0.11851306756337483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,768,0.09425066312154134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,2048,0.14473706881205242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,512,0.08980053265889486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,512,0.07996373176574707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,1024,0.09340266386667886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,256,0.06113919814427694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,256,0.0640394647916158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,1536,0.12307413419087727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,128,0.05225173234939575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,128,0.06409173409144084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,256,0.05591679811477661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,64,0.050058666865030924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,64,0.06092053254445394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,65536,32,0.04837119976679484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,65536,32,0.060628267129262295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,128,0.04928106864293416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,65536,4.2629547119140625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,65536,2.216688028971354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,16384,0.9807818730672201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,16384,0.4481877326965332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,512,0.07035733064015706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,12288,0.5116799990336101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,12288,0.3674933433532715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,65536,768,0.08036159674326579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,10240,0.3576383908589681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,10240,0.24301759401957193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,10240,0.15575893719991046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,8192,0.28754666646321614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,8192,0.274942938486735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,12288,0.18088000615437824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,7168,0.24950613975524902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,7168,0.24386986096700033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,16384,0.23251840273539223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,6144,0.21202774047851564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,6144,0.16801066398620607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,8192,0.12797973155975342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,5120,0.1786794662475586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,5120,0.13155946731567383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,7168,0.11370666821797688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,4096,0.14391892751057941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,4096,0.10605119864145915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,5120,0.08820160230000815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,3584,0.12678720156351725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,3584,0.10021119912465412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,4096,0.07242666880289714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,3072,0.10975360075632731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,3072,0.08587413628896078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,6144,0.10064960320790609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,2560,0.09169920285542807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,2560,0.0738602638244629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,2560,0.05134079853693644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,2048,0.074671999613444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,2048,0.06176533301671346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,3072,0.05894613265991211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,1536,0.0585375984509786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,3584,0.06604586839675904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,1536,0.05424213409423828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,1024,0.04146453142166138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,1024,0.04269653161366781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,1024,0.029717334111531574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,768,0.032415999968846636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,768,0.0378655989964803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,1536,0.03750720024108887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,512,0.02353066603342692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,512,0.033421866099039715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,768,0.026477867364883424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,256,0.016908800601959227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,256,0.02791573405265808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,256,0.01935360034306844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,128,0.01555519998073578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,128,0.030006400744120282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,128,0.01775040030479431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,64,0.016942934195200602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,64,0.03219733238220215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,16384,32,0.017794134219487508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,2048,0.044359465440114335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,512,0.023121066888173423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,16384,32,0.03209173281987508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,65536,1.5196661631266275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,16384,0.5433141072591146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,65536,3.3425771077473962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,16384,0.4009706815083821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,12288,0.3448437372843424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,12288,0.2736991882324219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,12288,0.1614847977956136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,10240,0.2913621266682943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,10240,0.21422826449076332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,16384,0.2086890697479248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,8192,0.230513064066569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,8192,0.17216320037841798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,10240,0.1379957358042399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,7168,0.20274880727132163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,7168,0.1495690663655599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,8192,0.11073280175526937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,6144,0.17149972915649414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,6144,0.12639040152231854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,7168,0.10046292940775554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,5120,0.14282132784525553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,5120,0.11172479788462322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,6144,0.08838613033294677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,4096,0.11296640237172444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,4096,0.08957653045654297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,5120,0.077619202931722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,3584,0.10038613478342692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,3584,0.08511892954508463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,4096,0.06401706536610921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,3072,0.0858901341756185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,3072,0.07348372936248779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,3584,0.05751573244730631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,2560,0.07345919609069824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,2560,0.062393601735432944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,3072,0.05122026602427164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,2048,0.05920960108439127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,2048,0.05351680119832357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,2560,0.044921600818634035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,1536,0.04601600170135498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,1536,0.04700160026550293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,1536,0.032181332508722946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,1024,0.03253973325093587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,1024,0.03701440095901489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,2048,0.038764798641204835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,768,0.02461120088895162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,768,0.0329802672068278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,1024,0.026012800137201947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,512,0.018994132677714028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,768,0.022737065951029457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,512,0.029528532425562543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,256,0.014127999544143677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,256,0.025281065702438356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,256,0.017383466164271034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,128,0.012330666184425354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,128,0.024037333329518636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,128,0.015820800264676412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,64,0.01015786627928416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,64,0.023036799828211465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,12288,32,0.010868266224861145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,512,0.02025066614151001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,12288,32,0.022975999116897582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,65536,1.177942403157552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,65536,2.8667147318522135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,16384,0.5493258794148763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,16384,0.3858570734659831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,16384,0.17634666760762532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,12288,0.2716416041056315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,12288,0.3190826733907064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,16384,65536,0.9101845423380533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,10240,0.3089162508646647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,10240,0.21856746673583985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,12288,0.13560746510823568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,8192,0.19969174067179363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,8192,0.14968640009562176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,10240,0.1165728012720744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,7168,0.17896745999654134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,7168,0.13677013715108235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,8192,0.09533332983652751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,6144,0.14788692792256672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,6144,0.11524906953175862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,12288,65536,0.7904341379801433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,5120,0.1239360014597575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,5120,0.10142292976379394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,6144,0.07639573415120443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,4096,0.09876266320546469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,4096,0.08246933619181315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,7168,0.08631680011749268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,3584,0.08949440320332845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,3584,0.07950293223063151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,5120,0.066594131787618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,3072,0.07606720129648845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,3072,0.06906773249308268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,3584,0.04967360099156697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,2560,0.06570560137430827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,2560,0.058968532085418704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,4096,0.05520960092544556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,2048,0.0526367982228597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,2048,0.04958186546961467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,3072,0.04483199914296468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,1536,0.0412778655687968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,1536,0.043569068113962814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,2048,0.03394560019175212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,1024,0.028705066442489623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,2560,0.039434667428334555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,1024,0.034777601559956864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,768,0.02327893376350403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,768,0.031074132521947222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,1536,0.028638933102289838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,512,0.01708266735076904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,512,0.027821866671244304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,512,0.018371200561523436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,256,0.012105600039164225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,256,0.02289173404375712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,768,0.02067199945449829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,128,0.01001706620057424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,1024,0.022565333048502605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,128,0.022205867369969687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,64,0.009165866176287334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,64,0.02226240038871765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,10240,32,0.009597866733868917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,10240,32,0.02132479945818583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,256,0.015369600057601929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,128,0.014949333667755128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,65536,1.0104042689005532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,65536,2.0553311665852863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,16384,0.363484795888265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,16384,0.3157855987548828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,12288,0.22866986592610677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,12288,0.20245013236999512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,12288,0.11174720128377277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,10240,0.1954975922902425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,10240,0.16522560119628907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,16384,0.14406506220499676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,8192,0.157805871963501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,8192,0.13007466793060302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,10240,0.09635519981384277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,7168,0.1381951967875163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,7168,0.1184618631998698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,8192,0.07915840148925782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,6144,0.11954666773478191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,6144,0.10251839955647786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,7168,0.07205866972605388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,5120,0.09896000226338705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,5120,0.09000746409098306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,6144,0.06366399923960367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,4096,0.08037439982096353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,4096,0.07124266624450684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,5120,0.055954134464263915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,3584,0.06998399893442789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,3584,0.0666752020517985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,10240,65536,0.6641183853149414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,3072,0.06163093249003092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,3072,0.061622401078542076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,4096,0.046476801236470536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,2560,0.05190293391545614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,2560,0.0487391988436381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,3584,0.04188906749089559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,2048,0.04193173249562581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,2048,0.04443413416544596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,3072,0.03783786694208781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,1536,0.03282133340835571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,1536,0.03761066595713298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,2560,0.033217066526412965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,1024,0.02434239983558655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,1024,0.03308159907658895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,1536,0.024394667148590087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,768,0.018711467583974205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,768,0.0308351993560791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,1024,0.019911466042200725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,512,0.014915200074513755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,512,0.026885332663853963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,2048,0.028173865874608357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,256,0.0108842670917511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,256,0.024482132991154988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,256,0.013661866386731466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,128,0.008796800176302593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,128,0.022644267479578654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,128,0.013171199957529703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,512,0.015887999534606935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,64,0.009186133742332458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,8192,32,0.00955733358860016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,64,0.02241493264834086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,8192,32,0.021715199947357176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,768,0.01785279909769694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,65536,0.9062933603922525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,65536,1.1956031799316407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,16384,0.32258453369140627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,16384,0.3182506561279297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,12288,0.18746986389160156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,12288,0.19621334075927735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,12288,0.08906559944152832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,10240,0.15607786178588867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,10240,0.13563733100891112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,16384,0.11405866940816242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,8192,0.1252949317296346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,8192,0.11083626747131348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,10240,0.07777600288391114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,7168,0.11011839707692464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,7168,0.10097493330637615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,8192,0.06377706527709961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,6144,0.09368639787038167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,6144,0.08664853572845459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,7168,0.05798293352127075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,5120,0.07999680042266846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,5120,0.07658560276031494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,6144,0.05174613396326701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,4096,0.06357760032018026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,4096,0.06170773506164551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,5120,0.04576106468836467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,3584,0.05709439913431803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,3584,0.057374934355417885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,4096,0.03752533197402954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,3072,0.04909439881642659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,3072,0.05124160051345825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,3584,0.03482773303985596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,2560,0.042173866430918375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,2560,0.04453226725260417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,3072,0.03076266646385193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,2048,0.033838931719462076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,2048,0.03824426730473836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,2560,0.027614933252334595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,1536,0.026557866732279462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,1536,0.03506453434626262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,2048,0.023324799537658692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,1024,0.018623999754587808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,1024,0.028787199656168622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,1536,0.02039360006650289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,768,0.015449600418408713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,768,0.026662399371465046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,768,0.015030399958292643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,512,0.011315199732780456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,512,0.024643200635910033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,1024,0.016237866878509522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,256,0.00819413314263026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,256,0.020602667331695558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,256,0.013617066542307535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,128,0.007124266525109608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,128,0.020572799444198608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,128,0.012851199507713318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,512,0.013829333583513894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,64,0.006286933521429698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,7168,32,0.006696533163388569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,32,0.018720000982284546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,7168,64,0.018679465850194296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,65536,0.8565216064453125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,65536,1.5422303517659506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,16384,0.31676692962646485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,16384,0.2634037335713705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,8192,65536,0.5378346761067708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,12288,0.17837759653727214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,12288,0.178274138768514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,12288,0.08920959631601968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,10240,0.14927147229512533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,10240,0.1345962683359782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,16384,0.11386559804280598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,8192,0.11900800069173176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,8192,0.10986986955006917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,10240,0.07781866391499838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,7168,0.10697173277537028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,7168,0.10024747053782146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,8192,0.0627669334411621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,6144,0.08913813432057699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,6144,0.08504106998443603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,7168,0.05769919951756795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,5120,0.07702933152516683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,5120,0.07263039747873942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,6144,0.05112959941228231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,4096,0.06034133434295654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,4096,0.05794133345286051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,5120,0.04468160072962443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,3584,0.05466560125350952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,3584,0.0543498675028483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,4096,0.03736213445663452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,3072,0.04797866741816203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,3072,0.04838826656341553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,3584,0.03414186636606852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,2560,0.039376000563303634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,2560,0.04182720184326172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,3072,0.030567467212677002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,2048,0.031089067459106445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,2048,0.03705173333485921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,2560,0.026936533053716023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,1536,0.024998400608698526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,1536,0.03328746755917867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,2048,0.022780799865722658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,1024,0.017364267508188883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,1024,0.02869759996732076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,1536,0.020052266120910645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,768,0.014454399545987448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,1024,0.016353066762288412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,768,0.027009065945943194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,512,0.012184533476829528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,512,0.02491626739501953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,512,0.013350400328636169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,256,0.009363200267155964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,256,0.02435200015703837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,768,0.014867200454076131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,128,0.009116799632708231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,128,0.022757333517074586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,128,0.012243200341860454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,64,0.008454400300979614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,256,0.012130133310953776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,64,0.022216532627741495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,6144,32,0.008785067001978557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,6144,32,0.020780799786249797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,65536,0.8059680302937826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,65536,0.833738644917806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,16384,0.2045952002207438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,16384,0.19621440569559734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,7168,65536,0.428656005859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,12288,0.15508586565653484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,12288,0.1646410624186198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,12288,0.09012479782104492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,10240,0.12603306770324707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,10240,0.11650559902191163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,16384,0.11507733662923177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,8192,0.1000106652577718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,8192,0.09569599628448486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,10240,0.07715840339660644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,7168,0.0863637367884318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,7168,0.08929173151652017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,8192,0.06307946840922038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,6144,0.07334613005320231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,6144,0.07445759773254394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,7168,0.05658239920934042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,5120,0.061868798732757566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,5120,0.06573119958241781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,6144,0.05056960185368856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,4096,0.049293867746988934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,4096,0.05247146685918173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,4096,0.037115732828776046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,3584,0.04358506600062052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,3584,0.04947946468989055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,5120,0.044044800599416095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,3072,0.0389631986618042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,3072,0.044819200038909913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,3584,0.034144000212351484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,2560,0.03228800098101298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,2560,0.040329599380493165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,3072,0.03020159999529521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,2048,0.02704319953918457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,2048,0.035323735078175864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,2560,0.026754132906595868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,1536,0.02104960083961487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,2048,0.022897066672643025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,1536,0.0317631999651591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,1024,0.015179733435312906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,1024,0.0271722674369812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,1024,0.016084266702334087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,768,0.012241066495577494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,768,0.027127466599146527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,1536,0.019513599077860513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,512,0.010609066486358643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,512,0.024716800451278685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,768,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,256,0.008283733328183492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,256,0.02202346722284953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,512,0.013372799754142762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,128,0.007549866537253062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,256,0.011962667107582092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,128,0.02030506730079651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,64,0.006791466474533081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,64,0.019638399283091225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,128,0.01120853324731191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,5120,32,0.007207466661930085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,5120,32,0.018569600582122803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,65536,0.5890261332194011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,65536,0.6853365580240885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,16384,0.15052693684895832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,16384,0.16855680147806804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,6144,65536,0.4169418652852376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,12288,0.11621440251668294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,12288,0.12555840015411376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,12288,0.06549013455708821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,10240,0.10083306630452474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,10240,0.10680639743804932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,16384,0.08416106700897216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,8192,0.08033599853515624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,8192,0.08732906977335611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,10240,0.05740586519241333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,7168,0.07167893250783285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,7168,0.07999040285746256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,8192,0.047353601455688475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,6144,0.060610131422678626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,6144,0.06624000072479248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,7168,0.04299519856770833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,5120,0.05160640080769857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,5120,0.056968533992767335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,6144,0.03841066757837931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,4096,0.04119040171305339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,4096,0.04750613371531169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,5120,0.033885868390401204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,3584,0.03704959948857625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,3584,0.04549013376235962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,4096,0.02898240089416504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,3072,0.031982932488123575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,3072,0.04121813376744588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,3584,0.02609493335088094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,2560,0.027884799242019653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,2560,0.03701653480529785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,3072,0.023483733336130776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,2048,0.02244266668955485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,2048,0.031385600566864014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,2560,0.020753065745035805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,1536,0.018268799781799315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,1536,0.02911679943402608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,1536,0.015837867061297098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,1024,0.013190399607022604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,1024,0.02540053327878316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,2048,0.018588799238204955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,768,0.010748799641927083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,768,0.022824533780415854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,1024,0.013316266735394797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,512,0.008097066481908163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,768,0.012403200070063274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,512,0.021306665738423665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,256,0.006369066735108693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,256,0.020551466941833497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,256,0.010877866546312969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,512,0.011240532994270325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,128,0.006004266440868378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,128,0.018911999464035035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,64,0.005880533158779145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,64,0.018898133436838785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,4096,32,0.006311466793219249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,4096,32,0.01838399966557821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,128,0.010428800185521444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,65536,0.5879797617594401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,65536,0.6751370747884115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,16384,0.14536426862080892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,16384,0.16589333216349283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,5120,65536,0.4285418510437012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,12288,0.11169919967651368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,12288,0.12474346955617269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,12288,0.06555946667989096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,10240,0.10182080268859864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,10240,0.1070911963780721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,16384,0.08383466402689616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,8192,0.08080426851908365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,8192,0.08678293228149414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,10240,0.05723093350728353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,7168,0.07120853265126546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,7168,0.0793386697769165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,8192,0.04697493314743042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,6144,0.06050453186035156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,6144,0.06533759832382202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,6144,0.037894399960835774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,5120,0.05134186744689941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,5120,0.056287999947865805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,7168,0.04268800020217896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,4096,0.04202026526133219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,4096,0.04756480058034261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,4096,0.028520532449086505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,3584,0.037309865156809494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,3584,0.04530239899953206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,5120,0.03310933311780294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,3072,0.03206506570180257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,3072,0.04110720157623291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,3584,0.02605760097503662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,3072,0.023487999041875204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,2560,0.027854933341344194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,2560,0.03686293363571167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,2048,0.022540799776713052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,2048,0.03298346598943074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,2048,0.018235733111699425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,1536,0.017811199029286705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,1536,0.02922240098317464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,2560,0.02062506675720215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,1024,0.013195733229319254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,1024,0.02505706747372945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,1024,0.013251200318336487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,768,0.010444800059000652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,768,0.022768000761667885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,1536,0.016084266702334087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,512,0.00790293316046397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,512,0.020588799317677816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,768,0.012446932991345723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,256,0.0058794667323430385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,512,0.01120746632417043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,256,0.018771199385325114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,128,0.005449600021044413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,128,0.01821440060933431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,4096,65536,0.3033162752787272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,128,0.010382933417956035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,256,0.010825600226720173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,64,0.004693333307902018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,64,0.018154666821161906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3584,32,0.005011199911435445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3584,32,0.016999467213948568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,65536,0.46099198659261065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,16384,0.11873280207316081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,65536,0.610371208190918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,16384,0.14974719683329266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,12288,0.08909013271331787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,12288,0.11040639877319336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,12288,0.06564906835556031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,10240,0.07630720138549804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,10240,0.09163626829783121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,16384,0.08403306802113851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,8192,0.06041599909464518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,8192,0.0744767983754476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,10240,0.057163735230763756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,7168,0.05292906761169434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,7168,0.06675733725229899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,8192,0.04650880098342895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,6144,0.04527039925257365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,6144,0.05538346767425537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,7168,0.0422048012415568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,5120,0.03823146820068359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,5120,0.04951146841049194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,6144,0.03791466554005941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,4096,0.03097813328107198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,4096,0.04193386634190877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,5120,0.033497599760691325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,4096,0.02797546585400899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,3584,0.027714133262634277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,3584,0.039869864781697587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,3072,0.02415786584218343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,3072,0.036892799536387126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,3072,0.02329813241958618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,2560,0.020907733837763467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,2560,0.03334826628367106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,3584,0.025890133778254193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,2048,0.017477333545684814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,2048,0.02892586588859558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,2048,0.01821333368619283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,2560,0.0209824005762736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,1536,0.013897599776585898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,1536,0.02684906721115112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,1024,0.009749333063761394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,1024,0.023982934157053628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,1024,0.01291306714216868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,768,0.008373333017031352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,768,0.022722133000691733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,1536,0.01566933294137319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,512,0.0068234667181968685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,512,0.022689066330591836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,768,0.01199679970741272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,256,0.005849599838256836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,512,0.011203199625015259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,256,0.01897066632906596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,128,0.005487999816735586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,128,0.018308266003926595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,128,0.01000853379567464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,256,0.010583466291427613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,64,0.005434666574001312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,3072,32,0.005437866846720377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,64,0.018182400862375894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,3072,32,0.018745599190394084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,65536,0.40172694524129227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,65536,0.5769941329956054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3584,65536,0.2988501230875651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,16384,0.10308799743652344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,16384,0.13986239433288575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,12288,0.0775328000386556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,12288,0.1028821309407552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,12288,0.06558826764424643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,10240,0.06327146689097086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,10240,0.08477973143259684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,16384,0.08417279720306396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,8192,0.05120533307393392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,8192,0.06813440322875977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,10240,0.056314667065938316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,7168,0.04322666724522908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,7168,0.05991679827372233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,8192,0.046469334761301676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,6144,0.03837759892145793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,6144,0.05156160195668539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,7168,0.04152213335037232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,5120,0.03168213367462158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,5120,0.04701120058695475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,6144,0.03736960093180339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,4096,0.026307199398676557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,4096,0.040299733479817704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,5120,0.03307733337084452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,3584,0.02296853264172872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,3584,0.03849813143412272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,4096,0.02798186739285787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,3072,0.02077546715736389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,3072,0.03520960013071696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,3584,0.025569067398707075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,2560,0.017565866311391197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,2560,0.031042132774988813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,2560,0.02063573400179545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,2048,0.014678399761517844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,2048,0.0273418664932251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,3072,0.023516800006230673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,1536,0.011668266852696736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,1536,0.026870399713516235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,1536,0.015758933623631795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,1024,0.00872320036093394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,1024,0.023294933636983237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,2048,0.018100267648696898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,768,0.007375999788443248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,768,0.02176106572151184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,768,0.012065066893895467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,1024,0.013054933150609335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,512,0.006233599781990051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,512,0.02097066640853882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,256,0.00558186670144399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,256,0.018202666441599527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,512,0.010823466380437215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,128,0.004632533093293508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,256,0.010545066992441813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,128,0.017934934298197428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,128,0.010075733065605164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,64,0.004340266684691111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,64,0.016645333170890807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2560,32,0.004705066482226053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2560,32,0.017543466885884602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,65536,0.30941759745279945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,65536,0.5306976000467937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,16384,0.07994346618652344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,16384,0.1274282693862915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,3072,65536,0.3015872001647949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,12288,0.061382400989532473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,12288,0.09357333183288574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,12288,0.042745598157246906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,10240,0.051651199658711754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,10240,0.07839039961496988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,16384,0.053405865033467614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,8192,0.04162559906641643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,8192,0.06169600089391073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,10240,0.037828266620635986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,7168,0.03766826788584392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,7168,0.05550080140431722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,8192,0.031023999055226643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,6144,0.03184746702512105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,6144,0.04872533480326335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,7168,0.029132799307505293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,5120,0.02760426600774129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,5120,0.04450026750564575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,6144,0.025862399737040204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,4096,0.02246613303820292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,4096,0.037113598982493085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,4096,0.019859200716018675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,3584,0.02014933427174886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,3584,0.036923734347025554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,5120,0.023541333278020223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,3072,0.018169599771499633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,3072,0.03312106728553772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,3584,0.018205867211023966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,2560,0.015482667088508605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,2560,0.030433066685994464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,3072,0.017091200749079386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,2048,0.013119999567667642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,2048,0.027154133717219038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,2560,0.014888532956441245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,1536,0.010365866621335347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,2048,0.01383680005868276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,1536,0.025434666872024538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,1024,0.007913599908351897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,1024,0.023101866245269775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,1024,0.010410666465759277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,768,0.006779733300209046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,768,0.020820266008377074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,1536,0.012128000458081562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,512,0.005890133480230967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,512,0.020798933506011964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,768,0.01034773290157318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,256,0.004638933142026265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,256,0.0187882661819458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,512,0.010072533289591472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,128,0.004373333354791006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,256,0.009556266665458679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,128,0.019048533837000527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,64,0.003977599988381068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,128,0.008736000458399455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,64,0.0184608002503713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,2048,32,0.004286933441956838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,2048,32,0.016695467631022136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,65536,0.24704106648763022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,65536,0.5052757263183594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2560,65536,0.3071370760599772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,16384,0.06607679923375448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,16384,0.1206602652867635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,12288,0.05187946557998657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,12288,0.08813440004984538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,12288,0.04288640022277832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,10240,0.04297600189844768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,10240,0.07121493021647135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,16384,0.05315946737925211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,8192,0.03421013355255127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,8192,0.056746665636698404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,10240,0.03680959939956665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,7168,0.02965013384819031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,7168,0.051381333669026696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,8192,0.03034026622772217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,6144,0.025595732529958087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,6144,0.04503680070241292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,7168,0.02805546720822652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,5120,0.02185279925664266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,5120,0.04198400179545085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,6144,0.025300266345342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,4096,0.018374399344126383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,4096,0.035062400499979655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,5120,0.023116799195607503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,3584,0.01625706652800242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,3584,0.033267199993133545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,4096,0.01991893251736959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,3072,0.014218667149543762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,3072,0.030946133534113567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,3584,0.018209065993626913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,3072,0.017056000232696534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,2560,0.012250666817029316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,2560,0.028677332401275634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,2048,0.010620799660682679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,2048,0.02502293388048808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,2560,0.015256533026695251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,1536,0.008577066659927367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,1536,0.02480106751124064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,2048,0.013700266679128012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,1024,0.006674133241176605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,1024,0.022627200682957968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,1536,0.01169599990049998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,768,0.006185600161552429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,768,0.021475199858347574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,1024,0.010431999961535137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,512,0.005166933437188466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,512,0.019330133994420372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,768,0.010752000411351522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,256,0.004276266694068909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,512,0.00950933297475179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,256,0.018293333053588868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,128,0.0038581334054470064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,128,0.016773333152135216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,256,0.009196799993515015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,64,0.0034282666941483817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,128,0.008760533730189006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,64,0.01687999963760376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1536,32,0.003836799909671148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1536,32,0.016167466839154564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,65536,0.16813333829243976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,65536,0.46401386260986327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,16384,0.04602239926656087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,2048,65536,0.186245330174764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,16384,0.1071786642074585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,12288,0.03426133394241333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,12288,0.07672213713328044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,12288,0.040856532255808514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,10240,0.027983999252319335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,10240,0.060850131511688235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,16384,0.05293973286946615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,8192,0.02284160057703654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,8192,0.04920639991760254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,10240,0.03582186698913574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,7168,0.020403200387954713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,7168,0.045723732312520346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,8192,0.029756800333658857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,6144,0.017901867628097534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,6144,0.04002666473388672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,7168,0.027614933252334595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,5120,0.015428266922632852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,5120,0.03710826635360718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,6144,0.025128533442815144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,4096,0.012902399897575379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,4096,0.03186666568120321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,5120,0.023130667209625245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,3584,0.011345066626866658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,3584,0.03101866642634074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,4096,0.019704532623291016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,3072,0.0100874662399292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,3072,0.02882453401883443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,3584,0.018194133043289186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,2560,0.008858666817347208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,2560,0.026784000794092815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,3072,0.01660480002562205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,2048,0.007570133109887441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,2048,0.024501333634058632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,2560,0.014918399850527444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,1536,0.006517333288987477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,1536,0.02410986622174581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,2048,0.013220266501108805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,1024,0.005109333495299021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,1024,0.020759467283884683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,1536,0.012009599804878235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,768,0.0046858668327331545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,1024,0.010735999544461567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,768,0.02054826617240906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,512,0.003819733361403147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,512,0.019040000438690186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,768,0.009943466385205586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,512,0.00993173321088155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,256,0.0034261333445707956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,256,0.018207999070485432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,128,0.0030069333811601003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,128,0.015334399541219077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,256,0.009130666653315227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,64,0.0029504001140594482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,64,0.014935466647148132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,128,0.00878613293170929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,1024,32,0.002994133283694585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,1024,32,0.014762666821479798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,65536,0.15211307207743327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,65536,0.4343594551086426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,16384,0.041416533788045246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,16384,0.10617492993672688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1536,65536,0.17954773902893068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,12288,0.030883200963338214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,12288,0.07345600128173828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,12288,0.04067519903182983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,10240,0.02614186604817708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,10240,0.05910933415095011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,16384,0.052693335215250645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,8192,0.022087466716766358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,8192,0.04858773152033488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,10240,0.035665067036946614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,7168,0.019822933276494346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,7168,0.0454037348429362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,8192,0.029665066798528032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,6144,0.01688746611277262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,6144,0.03909333149592082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,7168,0.027796266476313274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,5120,0.015145599842071533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,5120,0.037113598982493085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,6144,0.02516053318977356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,4096,0.01157973309357961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,4096,0.030988800525665283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,5120,0.022798933585484824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,3584,0.010855467120806376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,3584,0.031125332911809283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,4096,0.019782400131225585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,3072,0.009765332937240601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,3072,0.029025065898895263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,1024,65536,0.17841173807779948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,2560,0.008912000060081481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,2560,0.02709973255793254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,3584,0.01823893388112386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,2048,0.0075434664885203045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,2048,0.02482773264249166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,3072,0.01691306630770365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,1536,0.00626986672480901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,1536,0.022975999116897582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,2560,0.015341867009798685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,1024,0.005598933498064677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,1024,0.022348799308141074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,2048,0.013691733280817667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,768,0.0046965335806210835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,1536,0.01162559986114502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,768,0.020617600282033285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,512,0.0047082667549451195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,512,0.018681599696477254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,1024,0.010547199845314026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,768,0.009964799880981446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,256,0.0038090666135152185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,256,0.016521599888801575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,128,0.0034495999415715536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,512,0.009637332955996196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,256,0.009245866537094116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,128,0.01541866660118103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,64,0.0030847998956839246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,768,32,0.003454933315515518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,128,0.008936533331871032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,64,0.015010133385658264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,768,32,0.01567146678765615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,65536,0.0958079973856608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,16384,0.032909866174062094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,16384,0.10028586387634278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,65536,0.421556282043457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,12288,0.023042132457097374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,12288,0.06614400148391723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,12288,0.04063680171966553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,10240,0.01932906707127889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,10240,0.05357653299967448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,16384,0.05206826527913412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,8192,0.017666133244832356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,8192,0.045483732223510744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,10240,0.03532693386077881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,7168,0.01418346663316091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,7168,0.041679998238881424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,8192,0.029785599311192828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,6144,0.01283093293507894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,6144,0.03718719879786174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,7168,0.027564799785614012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,5120,0.010588799913724262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,5120,0.03502506812413533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,6144,0.025281065702438356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,4096,0.009317333499590557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,4096,0.030429865916570025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,5120,0.022564266125361124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,3584,0.00829013337691625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,3584,0.029806933800379437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,768,65536,0.1784053325653076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,3072,0.007893333335717519
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,3072,0.027825067440668743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,4096,0.019895466168721516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,2560,0.006780800223350525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,2560,0.025120000044504803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,3584,0.01813973387082418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,2048,0.0058794667323430385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,2048,0.022651733954747517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,3072,0.01656426688035329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,1536,0.005323733389377594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,2560,0.014722133676211039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,2048,0.013237333297729493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,1536,0.022758400440216063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,1024,0.0042453333735466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,1024,0.02040106654167175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,768,0.00395413339138031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,768,0.019661867618560792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,1024,0.010499200224876404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,1536,0.011723732948303223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,512,0.0033791999022165934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,512,0.018718934059143065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,768,0.010071466366449993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,256,0.0030346666773160298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,512,0.009640533725420635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,256,0.016566399733225504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,128,0.002647466709216436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,128,0.016039466857910155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,256,0.009191466371218364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,64,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,128,0.008694400389989216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,64,0.015225600202878317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,512,32,0.002603733291228612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,512,32,0.014497066537539164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,65536,0.06175253391265869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,65536,0.39755519231160485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,16384,0.016609066724777223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,16384,0.0959712028503418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,12288,0.016174933314323424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,12288,0.06185919841130575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,16384,0.05153706471125284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,10240,0.014442666371663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,10240,0.05130666494369507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,12288,0.039849599202473954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,8192,0.011729066570599873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,8192,0.04182399908701579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,10240,0.035640533765157065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,7168,0.010630399982134501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,8192,0.02960960070292155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,7168,0.04036053419113159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,6144,0.009646933277448018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,6144,0.03543039957682292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,7168,0.027586134274800618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,5120,0.008332799871762593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,5120,0.032927999893824264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,6144,0.02569813330968221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,4096,0.0071829333901405334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,4096,0.029029333591461183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,5120,0.02264960010846456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,3584,0.0067210664351781205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,3584,0.02884693344434102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,4096,0.019925334056218467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,3072,0.006237866481145223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,3072,0.026920533180236815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,3584,0.018133334318796792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,2560,0.00544106662273407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,2560,0.024807467063268026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,3072,0.017026132345199584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,2048,0.004747733473777771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,2560,0.01479680041472117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,2048,0.02294720013936361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,1536,0.00425493319829305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,512,65536,0.17887360254923504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,1536,0.022737065951029457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,1024,0.003532800078392029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,1024,0.020615466435750327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,2048,0.01362879971663157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,768,0.0032661333680152893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,1536,0.011627733707427979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,768,0.01888426740964254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,512,0.002993066608905792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,1024,0.010472533106803895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,512,0.018436266978581747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,768,0.010497066378593444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,256,0.0025781333446502686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,512,0.00958720048268636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,256,0.01681493322054545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,128,0.00264533335963885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,256,0.009181867043177288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,64,0.0022240000466505687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,128,0.015818666418393454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,128,0.009252267082532246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,256,32,0.002526933451493581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,64,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,256,32,0.015294933319091797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,65536,0.05292906761169434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,16384,0.011424000064531963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,16384,0.09141226609547934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,65536,0.39081707000732424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,12288,0.009586133559544881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,12288,0.05927679936091105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,256,65536,0.17930026054382325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,10240,0.010686933000882467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,10240,0.049643735090891525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,12288,0.03991146485010783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,8192,0.01128426690896352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,16384,0.05068480173746744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,8192,0.041713066895802814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,7168,0.010043733318646749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,7168,0.03916693528493245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,10240,0.03506666819254557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,6144,0.009217066566149394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,6144,0.03514346679051717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,8192,0.029665066798528032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,5120,0.007976533472537994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,7168,0.027639466524124145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,5120,0.033137067159016924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,4096,0.006904533505439759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,6144,0.02513813376426697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,4096,0.029069866736729937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,3584,0.006376533210277558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,5120,0.022638932863871256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,3584,0.028849067290623982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,4096,0.019419733683268228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,3072,0.005961599946022034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,3072,0.026811732848485308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,3584,0.018106667200724284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,2560,0.005094400048255921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,2560,0.024729599555333458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,3072,0.016702934106191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,2048,0.004743466774622599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,2560,0.014537599682807923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,2048,0.02283946673075358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,1536,0.0042357335488001505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,2048,0.0132341335217158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,1536,0.022533333301544188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,1024,0.0036469332873821257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,1536,0.011690666278203327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,1024,0.02068906625111898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,768,0.0031210665901501974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,1024,0.010351999600728353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,768,0.018968532482782997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,768,0.009616000453631084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,512,0.003009066730737686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,65536,0.18024214108784992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,512,0.017933867375055947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,256,0.0026261332134405774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,512,0.009614933530489604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,256,0.016824533541997276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,128,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,256,0.009231999516487122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,128,0.015995732943216958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,64,0.002345599979162216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2048,128,128,0.008408533533414204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,64,0.015027200182278952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,128,32,0.002699733277161916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,65536,0.04569386641184489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,128,32,0.014499200383822122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,16384,0.013588266571362815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,12288,0.011300266782442728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,16384,0.08951893647511801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,12288,0.05871893167495727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,10240,0.010217600067456563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,65536,0.3873813311258952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,8192,0.008728532989819845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,10240,0.04917226632436116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,7168,0.008371200164159138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,6144,0.007572266459465027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,7168,0.03908053239186605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,5120,0.007161599894364675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,8192,0.04300053517023723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,4096,0.006614399949709575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,6144,0.035046398639678955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,3584,0.006054399907588959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,5120,0.032943999767303465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,3072,0.005486933390299479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,4096,0.029155200719833373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,2560,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,3584,0.02882986664772034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,2048,0.004658133288224538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,3072,0.026434133450190227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,1536,0.0038794666528701783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,2048,0.02295253276824951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,1024,0.0033952000240484873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,2560,0.02509760061899821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,1536,0.0221834659576416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,768,0.0030016000072161358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,512,0.0026581334571043652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,1024,0.020435200134913126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,768,0.01944533387819926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,256,0.0026410666604836782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,512,0.018492799997329713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,128,0.0022805333137512207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,64,0.0021536000072956084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,256,0.017014400164286295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,64,32,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,128,0.014916266997655234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,32,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,64,64,0.015324800213177999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,65536,0.044842668374379474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,12288,0.008067200084527333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,16384,0.00956160028775533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,10240,0.007231999933719635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,12288,0.05907306671142578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,16384,0.09166080156962077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,10240,0.04937813282012939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,8192,0.01090666651725769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,7168,0.01046506663163503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,8192,0.04336426655451457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,65536,0.38698453903198243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,6144,0.009576533238093059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,7168,0.04209173520406087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,4096,0.007640533149242401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,5120,0.008389332890510559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,6144,0.035257601737976076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,3584,0.007142400244871776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,5120,0.03290026585261027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,3072,0.00544213354587555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,4096,0.028943999608357744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,2560,0.0050346667567888895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,3072,0.026899200677871705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,3584,0.028828799724578857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,2048,0.004217599829037985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,2560,0.024780799945195518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,1536,0.003818666686614355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,1024,0.0034282666941483817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,2048,0.023942399024963378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,1536,0.022312533855438233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,768,0.0030858665704727173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,1024,0.02005866765975952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,512,0.0026719999810059865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,768,0.018641066551208497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,256,0.0026517334083716077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,512,0.01813760002454122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,128,0.0023669332265853884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,256,0.01574399967988332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,64,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,128,0.01493013302485148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2048,32,32,0.002233600119749705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,64,0.014518400033315023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2048,32,32,0.013951999942461648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,16384,0.6794335683186848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,16384,2.115070978800456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,12288,0.9495157241821289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,12288,0.5097045262654623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,10240,1.0731136322021484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,10240,0.4408394813537598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,8192,0.819485855102539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,8192,0.6491221110026042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,8192,0.2387242635091146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,7168,0.5296223958333333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,7168,0.5708458582560222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,10240,0.290559991200765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,6144,0.4509546597798665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,6144,0.44653120040893557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,12288,0.34023574193318684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,5120,0.4425674756368001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,5120,0.2550677299499512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,6144,0.18366293907165526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,4096,0.30138454437255857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,4096,0.2634976069132487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,7168,0.211572265625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,3584,0.26458346048990883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,3584,0.1587509314219157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,5120,0.15844799677530924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,3072,0.2272490660349528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,3072,0.13984427452087403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,16384,0.44240748087565107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,2560,0.2178101380666097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,2560,0.141594664255778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,3584,0.11865706443786621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,2048,0.15073173840840656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,2048,0.10138879617055256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,4096,0.13092586994171143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,1536,0.1153760035832723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,1536,0.08419093290964762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,3072,0.10391360123952229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,1024,0.0812661329905192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,1024,0.06525333325068156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,2560,0.09080639680226645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,768,0.06424426635106405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,768,0.0559008002281189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,768,0.044354132811228436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,512,0.04816853205362956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,512,0.04779946804046631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,1536,0.06570880015691122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,256,0.03408106565475464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,256,0.044869331518809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,1024,0.05114880005518595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,128,0.02935146689414978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,128,0.039476267496744794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,128,0.028139734268188478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,64,0.026753065983454387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,64,0.039036798477172854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,65536,32,0.026757333676020307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,65536,32,0.03952213525772095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,65536,2.0566229502360027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,256,0.03123840093612671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,2048,0.07789973417917886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,65536,0.8438005447387695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,16384,0.30924053192138673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,16384,0.2183861255645752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,12288,0.2349631945292155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,12288,0.163918940226237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,65536,512,0.03858773310979207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,10240,0.1963029384613037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,10240,0.14175573984781903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,10240,0.09267306327819824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,8192,0.15884799957275392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,8192,0.11587626934051513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,12288,0.10864853064219157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,7168,0.13980906804402668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,7168,0.10761600335439045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,16384,0.14115413029988605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,6144,0.1208949327468872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,6144,0.09551146825154623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,8192,0.07551466623942057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,5120,0.1005621353785197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,5120,0.08024960358937581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,7168,0.06821866830190024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,4096,0.08626346588134766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,4096,0.06069653431574503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,5120,0.05284479856491089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,3584,0.07463359832763672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,3584,0.06315946578979492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,6144,0.06084479888280233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,3072,0.06546560128529867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,3072,0.05773013432820638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,4096,0.044283731778462725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,2560,0.05541333357493082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,2560,0.0511680006980896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,2560,0.03171093265215556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,2048,0.045389866828918456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,2048,0.043584001064300534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,3584,0.04043733278910319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,1536,0.03536320130030314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,1536,0.03560426632563273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,3072,0.03569600184758504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,1024,0.025425066550572712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,1024,0.03105066617329915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,1024,0.019138133525848387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,768,0.01911679903666178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,768,0.028626133998235066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,1536,0.023172267278035483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,512,0.0151936004559199
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,512,0.025543467203776045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,768,0.017156267166137697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,256,0.011539199948310852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,256,0.023923200368881226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,512,0.014510933558146158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,128,0.008745599786440532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,128,0.020866133769353232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,256,0.01288426617781321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,64,0.009317333499590557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,64,0.022194133202234904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,16384,32,0.009735467036565144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,16384,32,0.022813866535822548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,128,0.011963733037312825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,2048,0.0271999994913737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,65536,1.4109109242757163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,65536,0.7400938669840496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,16384,0.4601855913798015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,16384,0.22322346369425455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,12288,0.19598933855692546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,12288,0.17235520680745442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,12288,0.0842357317606608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,10240,0.14785706202189128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,10240,0.11559680302937825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,16384,0.10959146817525227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,8192,0.12005013624827068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,8192,0.0902026653289795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,10240,0.07212693691253662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,7168,0.10485440095265705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,7168,0.08438080151875814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,8192,0.059165867169698086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,6144,0.09119146664937337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,6144,0.07348159948984781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,7168,0.05404693285624186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,5120,0.07596159776051839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,5120,0.06364479859670004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,6144,0.048231466611226397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,4096,0.06055999994277954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,4096,0.053224531809488926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,5120,0.04172053337097168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,3584,0.0529749313990275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,3584,0.050253868103027344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,4096,0.035335465272267656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,3072,0.04598613182703654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,3072,0.045398398240407305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,3584,0.03218773404757182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,2560,0.038967466354370116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,2560,0.040831998984018965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,3072,0.028854399919509888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,2560,0.02542080084482829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,2048,0.031975466012954715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,2048,0.035214932759602864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,1536,0.02499306599299113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,1536,0.0308896005153656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,1536,0.0190720001856486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,1024,0.01768746574719747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,1024,0.027556266387303668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,2048,0.022193066279093423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,768,0.015319466590881348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,768,0.027110399802525838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,1024,0.015313067038853965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,512,0.011806933085123698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,512,0.022851200898488362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,768,0.013819733262062072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,256,0.009739733735720317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,512,0.012446932991345723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,256,0.022750933965047203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,128,0.009250133236249288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,128,0.020871466398239134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,128,0.01206826666990916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,64,0.008750933408737182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,64,0.021208532651265464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,12288,32,0.008849066495895386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,256,0.010782933235168457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,12288,32,0.022985599438349404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,65536,0.6184576034545899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,65536,1.1294282277425132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,16384,0.2132160027821859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,16384,0.15567572911580402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,16384,0.10965546766916912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,12288,0.16167252858479816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,12288,0.12023893197377522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,12288,0.0850325345993042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,10240,0.13098773161570232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,10240,0.10241066614786784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,16384,65536,0.5201237360636394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,8192,0.11081813176472981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,8192,0.08396053314208984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,10240,0.0728821357091268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,7168,0.09242133299509683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,7168,0.07879467010498047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,8192,0.058448000748952234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,6144,0.08354773521423339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,6144,0.0684554656346639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,12288,65536,0.39935146967569984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,5120,0.0680351972579956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,5120,0.05910720030466715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,7168,0.05371306737263998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,4096,0.05297706524531046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,4096,0.0497162659962972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,6144,0.04740906556447347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,3584,0.04868586858113606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,3584,0.04575573205947876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,5120,0.041493332386016844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,3072,0.04157973527908325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,3072,0.04250239928563436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,3584,0.03194239934285482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,2560,0.035607465108235675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,2560,0.03799253304799398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,4096,0.034721068541208905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,2048,0.02882986664772034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,2048,0.033163734277089435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,3072,0.02863466739654541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,1536,0.023093332846959434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,1536,0.02999040087064107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,2560,0.025228800376256307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,1024,0.017095466454823814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,1024,0.024796799818674723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,2048,0.02150826652844747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,768,0.013299199938774108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,768,0.02362133264541626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,768,0.01325866679350535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,512,0.009947733084360758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,512,0.020588799317677816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,1024,0.015381333231925965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,256,0.0073642666141192125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,256,0.01948480010032654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,1536,0.018580265839894614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,128,0.006648533542950948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,128,0.016718933979670204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,256,0.010863999525705974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,64,0.005990399916966756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,64,0.01737066706021627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,10240,32,0.005942399799823761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,10240,32,0.018730666240056357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,512,0.012053333719571431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,128,0.010390399893124899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,65536,0.6062143961588542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,16384,0.16449599266052245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,65536,0.5113087972005208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,16384,0.16654613812764485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,12288,0.12951786518096925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,12288,0.10555413564046223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,12288,0.06079039971033732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,10240,0.10191573301951092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,10240,0.09021759827931722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,16384,0.0775871992111206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,8192,0.0811626672744751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,8192,0.07213013172149658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,10240,0.05199679931004843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,7168,0.07263147036234538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,7168,0.06806186834971109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,8192,0.04349973201751709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,6144,0.061810131867726645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,6144,0.0607968012491862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,7168,0.03928533395131429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,5120,0.05321386655171713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,5120,0.05114453236262003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,6144,0.034791465600331625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,4096,0.04215146700541179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,4096,0.04306559960047404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,10240,65536,0.4085728009541829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,3584,0.03843733469645182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,3584,0.04353919823964437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,5120,0.03096959988276164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,3072,0.03312106728553772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,3072,0.037808001041412354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,3584,0.024498132864634196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,2560,0.02841599980990092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,2560,0.03509013255437215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,4096,0.02630400061607361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,2048,0.023145600159962972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,2048,0.030933332443237305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,3072,0.022275199492772423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,1536,0.018641066551208497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,1536,0.027353600660959883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,2560,0.01965973377227783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,2048,0.017015467087427773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,1024,0.013532800475756326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,1024,0.024493867158889772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,768,0.011947733163833619
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,768,0.024759467442830405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,1536,0.015003732840220132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,512,0.008679466446240743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,512,0.02304960091908773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,1024,0.012059733271598816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,256,0.007253333429495494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,256,0.018632533152898152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,512,0.01049066682656606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,128,0.005947733422120413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,768,0.01165013313293457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,128,0.020512000719706217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,64,0.006346666812896728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,64,0.018653867642084758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,8192,32,0.0063381334145863845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,8192,32,0.018590933084487914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,128,0.009586133559544881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,65536,0.4974687894185384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,256,0.009585066636403402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,16384,0.12557973066965739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,65536,0.4535573323567708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,16384,0.1121077299118042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,12288,0.0943733294804891
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,12288,0.09486186504364014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,12288,0.0602133313814799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,10240,0.07961599826812744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,10240,0.0833845297495524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,16384,0.07749013106028238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,8192,0.06582933266957601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,8192,0.06909546852111817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,10240,0.05182826519012451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,7168,0.05769919951756795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,7168,0.06599893172581991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,8192,0.04264426628748576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,6144,0.05062400102615357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,6144,0.05681279897689819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,7168,0.03862186670303345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,5120,0.04761600097020467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,5120,0.04774293502171834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,6144,0.03476053476333618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,4096,0.03946133454640706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,4096,0.041075201829274495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,5120,0.030905600388844805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,3584,0.03461440006891887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,3584,0.03892693519592285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,4096,0.02646080056826274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,3072,0.03012479941050212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,3072,0.036798934141794845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,3584,0.02405653397242228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,2560,0.025412267446517943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,2560,0.03290560046831767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,3072,0.02158400019009908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,2048,0.02151040037473043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,2048,0.0300053338209788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,2560,0.019067732493082683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,1536,0.016536532839139303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,1536,0.026794666051864625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,1536,0.014604799946149192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,1024,0.012148267030715943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,1024,0.022738132874170938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,2048,0.01702293356259664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,768,0.010006399949391682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,768,0.022285866737365722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,1024,0.012027733524640401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,512,0.00690773328145345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,512,0.020438400904337566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,768,0.011366400122642516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,256,0.005570133527119955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,256,0.01936639944712321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,512,0.009944533308347065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,128,0.005039999882380167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,128,0.017077332735061644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,256,0.009675733248392741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,128,0.009123200178146362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,64,0.004779733220736186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,64,0.016509866714477538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,7168,32,0.004722133278846741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,7168,32,0.017189333836237587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,65536,0.47049814860026046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,65536,0.4392405192057292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,16384,0.1351744016011556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,16384,0.12368960380554199
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,8192,65536,0.28024959564208984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,12288,0.0921002705891927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,12288,0.08670933246612549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,16384,0.07721707026163736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,10240,0.07563947041829427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,10240,0.0741312026977539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,12288,0.060412800312042235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,8192,0.061858133474985755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,8192,0.06007466713587443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,10240,0.0515498677889506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,7168,0.05374826590220133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,7168,0.056012801329294835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,8192,0.042905600865681966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,6144,0.04681280056635539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,6144,0.049992533524831136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,7168,0.03839893341064453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,5120,0.03857386509577433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,5120,0.04312959909439087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,6144,0.03473066488901774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,4096,0.03189226587613424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,4096,0.03719040155410767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,5120,0.030661332607269286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,3584,0.028036266565322876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,3584,0.03628053267796834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,4096,0.02649173339207967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,3072,0.025040000677108765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,3072,0.03341333468755086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,3584,0.023560533920923867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,2560,0.0213045338789622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,2560,0.030933332443237305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,3072,0.021313067277272543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,2048,0.01835626761118571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,2048,0.028280532360076903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,2560,0.01865066687266032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,1536,0.014033066232999167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,1536,0.02472426692644755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,2048,0.016621866822242738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,1024,0.01088853379090627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,1024,0.022932267189025878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,7168,65536,0.2791221300760905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,768,0.009224533041318258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,768,0.02262293299039205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,1024,0.012140799562136333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,512,0.007035733262697856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,512,0.01996799906094869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,1536,0.014570666352907815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,256,0.005883733431498209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,768,0.010863999525705974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,256,0.020039467016855876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,128,0.005555200080076853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,128,0.01867413322130839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,128,0.00913706620534261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,64,0.005490133166313171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,512,0.010154666503270467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,64,0.016934400796890257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,6144,32,0.005490133166313171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,256,0.009540266791979472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,6144,32,0.01811306675275167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,65536,0.4029706637064616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,65536,0.4067541440327962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,16384,0.10447466373443604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,16384,0.10354133447011311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,12288,0.07996373176574707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,12288,0.08158613046010335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,12288,0.059330133597056064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,10240,0.06519253253936767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,10240,0.06803413232167563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,16384,0.07782719930013021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,8192,0.051620264848073326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,8192,0.055003734429677334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,10240,0.050680534044901526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,7168,0.045406933625539145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,7168,0.05028479894002279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,8192,0.042506666978200276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,6144,0.03920960028966268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,6144,0.044631465276082354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,7168,0.038118398189544676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,5120,0.03345813353856404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,5120,0.03915733496348063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,6144,0.034721068541208905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,4096,0.027291733026504516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,4096,0.03460586468378703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,5120,0.030014934142430623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,3584,0.024549333254496257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,3584,0.033207466204961136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,4096,0.02605760097503662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,3072,0.02095573345820109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,3072,0.03193706671396891
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,3584,0.023549866676330567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,2560,0.01834026575088501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,2560,0.029368533690770464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,2560,0.018677333990732826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,2048,0.015223466356595359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,2048,0.025214932362238568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,3072,0.02063573400179545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,1536,0.012274133165677388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,1536,0.025110399723052977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,1536,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,1024,0.009032533566157023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,1024,0.021452800432840983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,2048,0.01654293338457743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,768,0.007567999760309856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,768,0.021926399072011313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,1024,0.011612799763679505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,512,0.006236800054709116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,512,0.018594133853912353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,768,0.010815999905268351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,256,0.005182933310667674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,6144,65536,0.27842025756835936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,256,0.01686506668726603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,128,0.004648533463478088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,128,0.01772693395614624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,128,0.009133866429328919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,512,0.009938133756319682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,64,0.0042922665675481165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,64,0.017908267180124917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,256,0.00959999958674113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,5120,32,0.004640000065167745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,5120,32,0.016659200191497803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,65536,0.2910559972127279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,65536,0.34061441421508787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,16384,0.0755839983622233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,16384,0.08893226782480876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,12288,0.059165867169698086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,12288,0.06638400157292684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,12288,0.036532266934712725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,10240,0.049556267261505124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,10240,0.05786346594492594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,16384,0.046453332901000975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,8192,0.03998399972915649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,8192,0.04695680141448975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,10240,0.03192533254623413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,7168,0.03578240076700846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,7168,0.04457813501358032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,8192,0.027400533358256023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,6144,0.030823467175165813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,6144,0.040158931414286295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,7168,0.02512213389078776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,5120,0.026497066020965576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,5120,0.03621333440144857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,6144,0.022938666741053264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,4096,0.021860265731811525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,4096,0.0324127991994222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,5120,0.02057173252105713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,3584,0.01946880022684733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,3584,0.03191360036532084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,4096,0.01801813244819641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,3072,0.017384533087412515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,3072,0.029607466856638592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,3584,0.01646506687005361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,2560,0.015135999520619711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,2560,0.026840533812840777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,3072,0.014568533500035605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,2048,0.012759466965993246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,2048,0.025212800502777098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,2560,0.01325653294722239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,1536,0.010444800059000652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,1536,0.02304853399594625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,2048,0.012134400010108948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,1024,0.007720533510049183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,1024,0.021911466121673585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,1536,0.010787199934323628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,768,0.006702933212121327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,768,0.020397865772247316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,1024,0.00956160028775533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,512,0.005667200187842051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,512,0.018795732657114664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,768,0.00913279950618744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,256,0.004690133531888326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,256,0.016771199305852254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,256,0.008338133494059246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,128,0.0040853333969910945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,128,0.016593066851298015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,512,0.008721066514650981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,64,0.0038463999827702843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,64,0.01664426624774933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,4096,32,0.0038762666285037995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,4096,32,0.01664426624774933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,128,0.00787199983994166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,65536,0.30111894607543943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,65536,0.3649034818013509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,16384,0.07969813346862793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,16384,0.08999146620432535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,5120,65536,0.2826773325602213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,12288,0.059729067484537754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,12288,0.06870613098144532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,16384,0.045152000586191815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,10240,0.05213119983673096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,12288,0.03610453208287557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,10240,0.06040639877319336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,8192,0.04163519938786824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,8192,0.04819306532541911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,8192,0.026964267094930012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,7168,0.03724693457285563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,7168,0.045576532681783036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,10240,0.031549866994222006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,6144,0.03147413333257039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,6144,0.04141866763432821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,7168,0.025141332546869916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,5120,0.027825067440668743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,5120,0.037221332391103104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,6144,0.022383999824523926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,4096,0.02266133427619934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,4096,0.03243199984232585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,4096,0.017374932765960693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,3584,0.020350933074951172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,3584,0.03158613244692485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,5120,0.020472532510757445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,3072,0.01841920018196106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,3072,0.030293333530426025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,3584,0.01576640009880066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,2560,0.015520000457763672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,2560,0.027577600876490277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,3072,0.014990933736165366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,2048,0.013594667116800943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,2048,0.02547093431154887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,2560,0.013214932878812155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,1536,0.010588799913724262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,1536,0.02295893430709839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,4096,65536,0.16102293332417805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,1024,0.007942399879296621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,1024,0.02089280088742574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,1536,0.010805333654085796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,768,0.006774400174617767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,768,0.0187008003393809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,2048,0.012241066495577494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,512,0.005906133353710175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,512,0.018947199980417887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,1024,0.009186133742332458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,256,0.0043488000830014546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,256,0.016607999801635742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,768,0.009157333771387737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,128,0.0038751999537150065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,512,0.008834133545557659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,128,0.016316800316174825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,64,0.003398400048414866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,256,0.008116266628106435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3584,32,0.003479466587305069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,64,0.015166933337847391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3584,32,0.01479680041472117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,128,0.00795413355032603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,65536,0.22820266087849936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,16384,0.0603498657544454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,65536,0.3309119860331217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,16384,0.07671146392822266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,12288,0.04566826820373535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,12288,0.05955946842829386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,12288,0.03550080060958862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,10240,0.039025068283081055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,10240,0.051541332403818765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,16384,0.04459306796391805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,8192,0.03226773341496785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,8192,0.042344534397125246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,10240,0.03141546646753947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,7168,0.027811199426651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,7168,0.04108800093332927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,8192,0.02683519919713338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,6144,0.025106134017308553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,6144,0.037229867776234944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,7168,0.024397865931193034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,5120,0.020916267236073812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,5120,0.03320213357607524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,6144,0.022120533386866252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,4096,0.017847466468811034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,4096,0.030668799082438154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,5120,0.020054399967193604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,3584,0.01576533317565918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,3584,0.029445334275563555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,4096,0.017052799463272095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,3072,0.014391466975212097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,3072,0.02762346665064494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,3584,0.015756799777348836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,2560,0.012429866194725036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,2560,0.02593066692352295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,2560,0.01283093293507894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,2048,0.010428800185521444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,2048,0.023861332734425863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,3072,0.01437333325544993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,1536,0.008701866865158081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,1536,0.022745599349339805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,2048,0.01160533328851064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,1024,0.006691200037797292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,1536,0.010470400253931682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,1024,0.020987733205159505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,768,0.005921066800753275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,768,0.01885439952214559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,768,0.008726400136947633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,512,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3584,65536,0.15783146222432454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,512,0.01798293391863505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,256,0.004247466723124186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,256,0.017512534062067667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,1024,0.009578667084376017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,128,0.003909333298603693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,128,0.016355199615160625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,256,0.007950933277606964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,64,0.0038250667353471124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,64,0.015796266992886863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,3072,32,0.003803733239571253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,128,0.0075765331586201985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,512,0.008410666386286418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,3072,32,0.01703146696090698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,65536,0.19220693906148273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,16384,0.05095680157343546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,16384,0.07129066785176595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,65536,0.31127252578735354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,12288,0.03911679983139038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,12288,0.053719464937845865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,12288,0.03529173135757446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,10240,0.03315839966138204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,10240,0.047323731581370036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,16384,0.04456746578216553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,8192,0.026764800151189167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,8192,0.039230934778849286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,10240,0.030988800525665283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,7168,0.024382932980855306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,7168,0.039241600036621097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,8192,0.02644053300221761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,6144,0.020744532346725464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,6144,0.03510719935099284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,7168,0.02401919960975647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,5120,0.018598399559656777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,5120,0.032474666833877563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,6144,0.021590399742126464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,4096,0.015229866902033488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,4096,0.029710932572682695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,5120,0.01990613341331482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,3584,0.013479466239611307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,4096,0.01693120002746582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,3584,0.028318933645884198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,3072,0.012197333574295043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,3072,0.027127466599146527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,3072,0.014456533392270408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,2560,0.010717866818110149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,2560,0.024975999196370443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,3584,0.015684266885121666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,2048,0.009127466877301534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,2048,0.023142399390538533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,2048,0.011654399832089742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,1536,0.007398400207360585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,1536,0.022778666019439696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,2560,0.01288106640179952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,1024,0.005965866645177205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,1024,0.020645334323247274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,1024,0.009219200412432352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,768,0.005381333331267038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,768,0.01917866667111715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,1536,0.010645332932472228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,512,0.00451200008392334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,512,0.016953599452972413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,768,0.009096533060073853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,256,0.00386559988061587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,256,0.01652479966481527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,512,0.008339200417200725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,128,0.0034111998975276947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,128,0.01639253298441569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,128,0.007891199986139933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,256,0.007916800181070964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,64,0.0033770665526390077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2560,32,0.0034048000971476236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,64,0.01543786625067393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2560,32,0.016618667046229045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,65536,0.1665770689646403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,65536,0.2865674654642741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,16384,0.04445226589838664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,16384,0.0670794645945231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,3072,65536,0.15536425908406576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,12288,0.03445226748784383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,12288,0.049491198857625325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,12288,0.03510613441467285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,10240,0.028843732674916585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,10240,0.04488639831542969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,16384,0.04394986629486084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,8192,0.023285333315531412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,8192,0.03827306826909383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,10240,0.030938667058944703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,7168,0.020913066466649373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,7168,0.03710506757100423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,8192,0.025916800896326704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,6144,0.018309332927068076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,6144,0.033124266068140666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,7168,0.02393066684405009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,5120,0.015939199924468996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,5120,0.03070080081621806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,6144,0.021810134251912437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,4096,0.013154133160909017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,4096,0.027162667115529376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,5120,0.02018773357073466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,3584,0.012210133671760558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,3584,0.026811732848485308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,4096,0.017336533466974894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,3072,0.010755200187365215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,3072,0.02494293252627055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,3584,0.01613866686820984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,2560,0.009393067161242167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,2560,0.024372265736262003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,3072,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,2048,0.007389866809050243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,2048,0.02262506683667501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,2560,0.012836266557375589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,1536,0.006235733131567637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,1536,0.02147093415260315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,2048,0.01200320025285085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,1024,0.005062399804592133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,1024,0.0191210667292277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,1536,0.0107914666334788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,768,0.004625066618124644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,768,0.019729065895080566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,1024,0.009550933043162029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,512,0.003899733225504557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,512,0.01660160024960836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,768,0.008734933535257975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,256,0.0034005333979924522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,512,0.008285866677761078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,256,0.016745599110921223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,256,0.008321066697438557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,128,0.0030570665995279947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,128,0.016004266341527303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,128,0.00790719985961914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,64,0.0029994666576385496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,64,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,2048,32,0.0030389333764712016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,2048,32,0.01452906628449758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2560,65536,0.15211307207743327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,65536,0.1373696009318034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,65536,0.2557802677154541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,16384,0.037251198291778566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,16384,0.06025173266728719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,12288,0.029150933027267456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,12288,0.045551999409993486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,12288,0.03517760038375854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,10240,0.023718400796254476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,10240,0.0413045326868693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,16384,0.044044800599416095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,8192,0.019101866086324058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,8192,0.0353770653406779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,10240,0.03102506597836812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,7168,0.017795199155807497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,7168,0.035230934619903564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,8192,0.02555946707725525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,6144,0.015044266978899637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,6144,0.030909866094589233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,6144,0.02185493310292562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,5120,0.013179733355840047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,5120,0.028858667612075804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,7168,0.024027733008066814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,4096,0.011166933178901672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,4096,0.02584106723467509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,5120,0.0197653333346049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,3584,0.009724799791971843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,3584,0.02495786746342977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,4096,0.01699626644452413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,3072,0.008975999553998311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,3072,0.02492799957593282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,3584,0.015683199961980185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,2560,0.007531733314196269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,3072,0.014591999848683677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,2560,0.023769599199295045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,2048,0.006663466493288676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,2048,0.021516799926757812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,2048,0.01162453293800354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,1536,0.0058677335580190025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,2560,0.01307093302408854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,1536,0.02089280088742574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,1024,0.004699733356634776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,1024,0.018668800592422485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,1024,0.009165866176287334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,768,0.004215466479460398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,768,0.01848640044530233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,1536,0.010963199536005656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,512,0.003899733225504557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,512,0.016919465859731038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,768,0.008736000458399455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,256,0.00373333344856898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,256,0.016500266393025716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,512,0.008476799726486206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,128,0.0031626666585604347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,128,0.016123732924461363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,256,0.007926400005817413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,64,0.003014400104681651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,64,0.014519466956456503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1536,32,0.002997333308060964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,128,0.00768746683994929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1536,32,0.015290666619936624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,65536,0.09341973463694254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,65536,0.24306133588155113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,16384,0.030728532870610552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,16384,0.053541334470113125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,2048,65536,0.1518229325612386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,12288,0.023611734310785927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,12288,0.043933868408203125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,12288,0.034739200274149576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,10240,0.020422399044036865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,10240,0.04076693455378215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,16384,0.043212799231211345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,8192,0.016687999169031777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,8192,0.033266133069992064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,10240,0.03068266709645589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,7168,0.015378133455912272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,7168,0.03344959815343221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,8192,0.025526400407155352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,6144,0.012965333461761475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,6144,0.03006613254547119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,6144,0.021497599283854165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,5120,0.01125973363717397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,5120,0.027142399549484254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,7168,0.024099200963974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,4096,0.009227733810742695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,4096,0.02524906595547994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,5120,0.019754666090011596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,3584,0.008238933483759562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,4096,0.017053866386413576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,3584,0.02480213244756063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,3072,0.007592533528804779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,3072,0.024184532960255942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,3584,0.01577279965082804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,2560,0.007057066758473713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,2560,0.023117866118748984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,3072,0.014070399602254233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,2048,0.005998933315277099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,2048,0.021093332767486574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,2560,0.01288106640179952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,1536,0.005100800096988678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,1536,0.020595200856526694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,2048,0.01209173301855723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,1536,0.010785067081451416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,1024,0.00439573327700297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,1024,0.019006933768590292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,768,0.00383146678407987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,768,0.016821332772572837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,1024,0.009725866715113322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,768,0.00874773363272349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,512,0.0035317334036032355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,512,0.01697280009587606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,256,0.0034485332667827605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,256,0.01653866668542226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,512,0.008488532900810242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,256,0.00795413355032603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,128,0.0031221332649389905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,128,0.015943466623624166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,64,0.0029493334392706556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,64,0.014589866995811463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,1024,32,0.002917333443959554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,128,0.008010666569073994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,1024,32,0.01483626663684845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,65536,0.08208853403727213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,65536,0.22627199490865074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,16384,0.023896533250808715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,16384,0.05337280035018921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1536,65536,0.15133973757425945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,12288,0.018331732352574667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,12288,0.041317331790924075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,12288,0.034306132793426515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,10240,0.015612799922625223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,10240,0.03716586828231812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,16384,0.04314346710840861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,8192,0.012947199741999307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,8192,0.03164586623509725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,8192,0.025950932502746583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,7168,0.011558399597803751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,7168,0.031005867322285968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,10240,0.030542933940887453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,6144,0.01055573324362437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,6144,0.02899199922879537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,7168,0.023499733209609984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,5120,0.008829866846402485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,5120,0.026818132400512694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,6144,0.021489065885543824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,4096,0.007776000102361043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,4096,0.025538132588068647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,1024,65536,0.1516053358713786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,3584,0.007266133526961009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,3584,0.024679466088612875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,5120,0.01957013408342997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,3072,0.006296533346176148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,3072,0.023112533489863078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,4096,0.016962132851282754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,2560,0.0059125334024429325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,2560,0.022954666614532472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,3584,0.01618026693662008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,2048,0.005117866893609365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,2048,0.02074880003929138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,3072,0.014033066232999167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,1536,0.00462719996770223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,1536,0.020627200603485107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,2560,0.012892799576123557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,1024,0.0038624001046021783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,1024,0.01885653336842855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,1536,0.010434133807818095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,2048,0.012104533116022746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,768,0.0034175999462604523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,768,0.017947733402252197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,1024,0.009316266576449076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,512,0.0030805334448814393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,512,0.016768000523249307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,256,0.003050666550795237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,768,0.008733866612116496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,256,0.016076800227165223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,512,0.008470400174458822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,128,0.0027295999228954316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,256,0.008317866424719492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,128,0.014789332946141561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,64,0.0026229334374268847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,768,32,0.0026752000053723653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,128,0.00796693315108617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,64,0.01513813336690267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,768,32,0.015352533260981242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,65536,0.05280960003534953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,16384,0.01801919937133789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,65536,0.22047999699910484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,16384,0.04984213511149089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,12288,0.016307199994723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,12288,0.03970453341801961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,12288,0.03415466547012329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,10240,0.014261333147684732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,16384,0.04341439803441365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,10240,0.037357866764068604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,8192,0.01248426636060079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,8192,0.03125653266906738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,10240,0.030573866764704388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,7168,0.010865066448847454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,7168,0.03091946641604106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,8192,0.0256223996480306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,6144,0.009352533022562663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,6144,0.028861866394678755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,7168,0.023941334088643393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,5120,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,5120,0.027386667331059773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,6144,0.0218560000260671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,4096,0.007124266525109608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,5120,0.01984106699625651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,4096,0.02337813377380371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,3584,0.00645653357108434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,3584,0.023885866006215416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,3584,0.015729066729545594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,4096,0.017374932765960693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,3072,0.005921066800753275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,3072,0.024579199155171712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,2560,0.005545599758625031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,2560,0.023778132597605386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,768,65536,0.15134933789571126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,2048,0.004759466648101807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,2048,0.021792000532150267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,3072,0.014039466778437296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,1536,0.0042581334710121155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,2560,0.012842667102813721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,1536,0.019157334168752035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,1024,0.003419733295838038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,1024,0.017973333597183228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,2048,0.01202133297920227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,768,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,768,0.01816426714261373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,1536,0.010761599739392598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,512,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,1024,0.009193600217501322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,512,0.016642133394877114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,256,0.0026026666164398193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,768,0.008776533603668212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,256,0.014896000425020853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,512,0.008726400136947633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,128,0.002540799975395203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,256,0.008309333523114523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,128,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,64,0.0022474666436513263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,128,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,512,32,0.002488533407449722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,64,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,512,32,0.0144405335187912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,65536,0.03609493176142375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,16384,0.012074666221936543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,65536,0.20478612581888833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,16384,0.04591360092163086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,12288,0.00997759997844696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,12288,0.03922453324000041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,12288,0.03461546500523885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,10240,0.010340266426404317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,10240,0.036849065621693926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,16384,0.04285973310470581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,8192,0.01120853324731191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,8192,0.031165866057078044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,10240,0.030081067482630414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,7168,0.010361599922180175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,8192,0.025941334168116253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,7168,0.03096959988276164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,6144,0.009100799759229023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,6144,0.02885546684265137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,7168,0.023589332898457847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,5120,0.007886933286984761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,5120,0.026392533381779985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,6144,0.021514666080474854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,4096,0.006750933329264323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,512,65536,0.1518837292989095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,4096,0.024215465784072875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,3584,0.00625600020090739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,3584,0.02431466579437256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,5120,0.019427200158437095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,3072,0.00557226687669754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,3072,0.022705066204071044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,4096,0.01694399913152059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,2560,0.005142400165398916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,3584,0.015773866573969522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,2560,0.022282665967941283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,2048,0.004661333560943603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,2048,0.02063360015551249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,3072,0.014072533448537192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,1536,0.003903999924659729
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,2560,0.012870400150616964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,1536,0.01911679903666178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,1024,0.003419733295838038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,1024,0.01887893279393514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,2048,0.01209279994169871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,768,0.003011200080315272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,1536,0.010418132940928141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,1024,0.009197866916656494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,768,0.017174400885899863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,512,0.0026719999810059865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,512,0.016500266393025716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,768,0.009105066458384197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,256,0.0026911998788515727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,512,0.008332799871762593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,256,0.015105066696802774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,128,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,128,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,256,0.007899733384450276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,64,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,64,0.014003200332323709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,128,0.00786240001519521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,256,32,0.0022143999735514323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,256,32,0.014598400394121806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,65536,0.028128000100453694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,65536,0.2008128007253011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,16384,0.009542399644851684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,16384,0.04536106586456299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,16384,0.04282879829406738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,12288,0.00803413341442744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,12288,0.037828266620635986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,256,65536,0.1526026725769043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,10240,0.01004266639550527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,10240,0.03522773186365764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,8192,0.008819199601809184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,8192,0.03140693306922913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,12288,0.03432426850001018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,7168,0.008402132987976074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,10240,0.030133332808812457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,7168,0.03094186584154765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,6144,0.007628799974918365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,8192,0.025961599747339886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,6144,0.029186133543650312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,5120,0.007154133419195812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,5120,0.026828799645106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,7168,0.02370986739794413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,4096,0.006571733454863231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,6144,0.021524266401926676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,4096,0.023316266139348348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,5120,0.019800533850987755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,3584,0.00592853327592214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,3584,0.024216532707214355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,3072,0.005560533205668131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,4096,0.016924800475438435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,3072,0.0236245334148407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,2560,0.0050570666790008545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,3584,0.01576533317565918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,2560,0.022541866699854533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,2048,0.004757333298524221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,3072,0.014338133732477823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,2048,0.020911999543507895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,1536,0.0039264000952243805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,2560,0.012843733032544455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,1536,0.01971199909845988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,2048,0.012117333213488261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,1024,0.0034485332667827605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,1536,0.010424533486366272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,1024,0.01866133411725362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,768,0.0030080000559488933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,65536,0.1516981283823649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,1024,0.009293867150942485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,768,0.017358932892481485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,512,0.002632533262173335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,256,0.002630399912595749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,512,0.016607999801635742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,768,0.008734933535257975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,512,0.008328533172607422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,128,0.002170666555563609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,256,0.01579200029373169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,256,0.007901866734027863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,64,0.0022122666239738466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,128,0.015477333466211954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,128,32,0.0022111999491850535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,64,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1024,128,128,0.007480533421039581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,65536,0.02876586715380351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,128,32,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,16384,0.008725333213806152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,12288,0.007797333101431529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,10240,0.007095466554164887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,16384,0.04253546794255574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,12288,0.03663680156071981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,8192,0.0071050668756167095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,65536,0.2025983969370524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,7168,0.006723199784755707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,10240,0.034953598181406656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,6144,0.006263466676076253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,8192,0.03100053270657857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,7168,0.030937600135803222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,5120,0.006663466493288676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,6144,0.02874773343404134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,4096,0.005922133227189382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,3584,0.005966933568318685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,5120,0.026919466257095338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,4096,0.022674133380254112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,3072,0.005460266768932342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,3584,0.024528000752131143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,2560,0.005050666630268097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,2048,0.004284800092379252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,1536,0.0038634667793909705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,2560,0.02137920061747233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,3072,0.022775467236836752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,2048,0.020572799444198608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,1024,0.0030730667213598887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,1536,0.0195360004901886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,768,0.0029663999875386557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,1024,0.018323200941085815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,512,0.002609066665172577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,256,0.0025578667720158894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,768,0.016721065839131674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,512,0.01655893325805664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,128,0.0022069332500298817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,256,0.014965333541234336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,64,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,64,32,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,128,0.014751999576886495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,64,0.01402666668097178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,65536,0.023385600248972575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,64,32,0.014631467064221701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,16384,0.010378666718800863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,12288,0.008901333808898926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,16384,0.04327893257141113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,65536,0.19788479804992676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,10240,0.008400000135103862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,12288,0.03678826491038005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,8192,0.00757120003302892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,7168,0.007107200225194295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,10240,0.034924801190694174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,8192,0.030874667565027873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,6144,0.00668693333864212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,5120,0.0062047998110453285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,7168,0.030937600135803222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,6144,0.028576000531514482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,4096,0.0058794667323430385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,5120,0.02601813276608785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,3584,0.0055754666527112326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,3072,0.005479466418425242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,4096,0.024270933866500855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,3584,0.024010666211446128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,2560,0.005062399804592133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,2048,0.005138133466243744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,3072,0.022824533780415854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,2560,0.022398932774861654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,1536,0.004470400015513102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,1024,0.003638399889071783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,2048,0.02101226647694906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,1536,0.019387733936309815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,768,0.003078400095303853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,512,0.0029578665892283124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,1024,0.018589866161346436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,768,0.01730133295059204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,256,0.0026026666164398193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,128,0.0022240000466505687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,512,0.01684053341547648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,256,0.01611840029557546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,64,0.0022677332162857056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1024,32,32,0.0021749332547187803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,128,0.015784533818562825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,64,0.01411733329296112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1024,32,32,0.015281066298484802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,16384,0.5221151987711589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,16384,1.2480032602945963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,12288,0.40124053955078126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,12288,1.0697556813557942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,10240,0.6052586873372395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,10240,0.34013439814249674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,8192,0.593885866800944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,8192,0.27581014633178713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,8192,0.18980693817138672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,7168,0.4975263913472493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,7168,0.3483509381612142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,10240,0.23599467277526856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,6144,0.368283748626709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,6144,0.21002666155497232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,12288,0.27639573415120444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,5120,0.3117386817932129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,5120,0.18034133911132813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,6144,0.14910079638163248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,4096,0.24640746116638185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,4096,0.1503712018330892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,7168,0.16900053024291992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,3584,0.2132213274637858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,3584,0.1431370735168457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,5120,0.12710293134053546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,3072,0.185533873240153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,3072,0.13205546538035076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,16384,0.3602527936299642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,2560,0.15431253115336102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,2560,0.11033706665039063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,3584,0.09485867023468017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,2048,0.12387946446736653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,2048,0.09312427043914795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,4096,0.10564906597137451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,1536,0.10238293011983235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,1536,0.07712746461232503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,3072,0.08362987041473388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,1024,0.07334613005320231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,1024,0.057171201705932616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,2560,0.07353706359863281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,768,0.058390398820241295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,768,0.05101653337478638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,2048,0.06315093437830607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,512,0.041766401131947836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,512,0.04543360074361165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,768,0.035995733737945554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,256,0.031185066699981688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,256,0.04097386598587036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,1024,0.04092586835225423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,128,0.021953066190083824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,128,0.03368639945983887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,128,0.023316266139348348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,64,0.019772799809773763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,64,0.032637866338094075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,65536,32,0.019078399737675986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,65536,32,0.03322880069414775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,1536,0.05233386754989624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,65536,1.054579226175944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,65536,0.6468479792277019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,16384,0.24333225886027016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,16384,0.17456639607747396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,256,0.025656533241271973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,12288,0.1836842695871989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,12288,0.13132800261179606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,65536,512,0.031524266799290976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,10240,0.15906559626261393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,10240,0.1063050667444865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,10240,0.07085013389587402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,8192,0.13009813626607258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,8192,0.08941120306650797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,12288,0.08226346969604492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,7168,0.1145525296529134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,7168,0.08017173608144125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,16384,0.10735999743143718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,6144,0.10019199848175049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,6144,0.06984426975250244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,8192,0.05797013441721598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,5120,0.0770464022954305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,5120,0.06160639921824137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,7168,0.0524895985921224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,4096,0.062326399485270176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,4096,0.055105066299438475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,5120,0.0405781348546346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,3584,0.0541536013285319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,3584,0.05046506722768148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,4096,0.03485333522160848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,3072,0.047915732860565184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,3072,0.04562453428904216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,6144,0.04671786626180013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,2560,0.04000213146209717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,2560,0.03996266523996989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,3072,0.028413865963617963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,2048,0.03308373292287191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,2048,0.035282135009765625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,2560,0.025282132625579833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,1536,0.02537173430124919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,1536,0.030972800652186078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,3584,0.031174399455388385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,1024,0.018057600657145182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,1024,0.02791786591211955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,1024,0.015410133202870688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,768,0.014615466197331747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,768,0.025212800502777098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,1536,0.01914773384730021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,512,0.012376532951990763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,512,0.022940800587336222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,768,0.013384532928466798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,256,0.009529599547386169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,256,0.02313386599222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,256,0.010732799768447876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,128,0.009099733829498292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,128,0.02148266633351644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,2048,0.022150399287541707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,64,0.008866133292516072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,128,0.010424533486366272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,64,0.02082879940668742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,16384,32,0.008709333340326945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,16384,32,0.02071466644605001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,512,0.012131200234095255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,65536,0.7142933527628581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,65536,0.48929494222005204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,16384,0.12585279941558838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,16384,0.3676266670227051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,12288,0.2057919979095459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,12288,0.14688000679016114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,12288,0.058942933877309166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,10240,0.11422293186187744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,10240,0.0901151974995931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,16384,0.07662399609883627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,8192,0.09440106550852458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,8192,0.07613440354665121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,10240,0.051319468021392825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,7168,0.08089599609375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,7168,0.06750826835632324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,8192,0.04263893365859985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,6144,0.07167147000630697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,6144,0.058745598793029784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,7168,0.038718934853871664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,5120,0.059476268291473386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,5120,0.05297173261642456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,6144,0.034713598092397054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,4096,0.04953813155492147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,4096,0.047653333346048994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,5120,0.030548266569773358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,3584,0.042080001036326094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,3584,0.042555733521779375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,4096,0.025832533836364746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,3072,0.037071998914082846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,3072,0.03914026816685994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,3584,0.02378986676534017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,2560,0.03088639974594116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,2560,0.034069331487019856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,3072,0.021329067150751748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,2048,0.026309333244959515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,2048,0.031677865982055665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,2560,0.01944533387819926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,1536,0.01975359916687012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,1536,0.027129600445429485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,1536,0.015310933192571005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,1024,0.014673067132631936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,1024,0.023178666830062866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,2048,0.017272533973058064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,768,0.012199466427167256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,768,0.02104533314704895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,1024,0.01176533301671346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,512,0.009378133217493693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,512,0.020146133502324422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,768,0.010808533430099488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,256,0.006309333443641663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,256,0.019469867149988808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,512,0.009958400328954061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,128,0.005875200033187866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,128,0.01885120073954264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,256,0.009239466985066731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,64,0.005382399757703145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,128,0.009795199831326802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,64,0.018551466862360637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,12288,32,0.005471999943256378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,12288,32,0.01722666621208191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,65536,0.5998218536376954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,65536,0.45110505421956376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,16384,0.14966293970743816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,16384,0.1201738675435384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,16384,0.07554986476898193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,12288,0.11291200319925945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,12288,0.09289066791534424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,12288,0.05816746552785238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,10240,0.09940586884816488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,10240,0.0780682643254598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,16384,65536,0.3989599863688151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,8192,0.08202346960703531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,8192,0.07666026751200358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,10240,0.05046506722768148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,7168,0.06979733308156331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,7168,0.06001493136088053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,8192,0.04155626694361369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,6144,0.060728534062703454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,6144,0.05330026547114054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,12288,65536,0.28052692413330077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,5120,0.049131735165913897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,5120,0.04945493141810099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,6144,0.033572268486022946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,4096,0.04042453368504842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,4096,0.04089920123418172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,7168,0.03818560043970744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,3584,0.03569600184758504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,3584,0.038496001561482744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,3584,0.023564799626668295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,3072,0.030842665831247968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,3072,0.03497920036315918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,4096,0.025329067309697466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,2560,0.026459733645121258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,2560,0.031956267356872556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,5120,0.029981867472330732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,2048,0.02177493373552958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,2048,0.029226666688919066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,3072,0.020984532435735066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,1536,0.017147733767827352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,1536,0.026361600557963057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,2560,0.01922986706097921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,1024,0.012956800063451132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,1024,0.023362133900324503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,2048,0.01708266735076904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,768,0.01050453285376231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,768,0.0217631995677948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,768,0.01120853324731191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,512,0.008661333719889324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,512,0.021862399578094483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,1024,0.01160533328851064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,256,0.006704000135262807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,256,0.02027413249015808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,1536,0.014454399545987448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,128,0.00551146666208903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,128,0.018617600202560425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,256,0.00918506681919098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,64,0.006281599899133046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,64,0.017732266585032144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,10240,32,0.006321066617965698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,10240,32,0.018518400192260743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,128,0.00872213343779246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,65536,0.4759861310323079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,512,0.009959466258684794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,65536,0.3926517486572266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,16384,0.12051946322123211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,16384,0.10595946311950684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,12288,0.09233600298563639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,12288,0.08303573131561279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,12288,0.05762133200963339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,10240,0.07825813293457032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,10240,0.06994667053222656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,16384,0.07469866275787354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,8192,0.062251734733581546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,8192,0.058957866827646886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,10240,0.04991893370946248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,7168,0.05554879903793335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,7168,0.0534229318300883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,8192,0.04155093431472778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,6144,0.04722239971160889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,6144,0.047288533051808676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,7168,0.03749119838078817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,5120,0.039953064918518064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,5120,0.04111466805140178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,10240,65536,0.27123839060465493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,4096,0.03205440044403076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,4096,0.0374069333076477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,6144,0.033565866947174075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,3584,0.029154133796691895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,3584,0.035361067454020186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,5120,0.029470932483673096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,3072,0.025541333357493083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,3072,0.03271039923032125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,3584,0.023231999079386393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,2560,0.021660800774892172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,2560,0.030689066648483275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,4096,0.02524799903233846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,2048,0.01775146722793579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,2048,0.02759893337885539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,3072,0.02121280034383138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,1536,0.015060266852378846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,1536,0.02488320072491964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,2560,0.01892906626065572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,1024,0.011352533102035522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,1024,0.020807466904322305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,2048,0.01663253307342529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,768,0.008898133039474487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,768,0.02067306637763977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,1024,0.011756799618403117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,512,0.0071829333901405334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,512,0.018973867098490395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,1536,0.01432960033416748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,256,0.00602346658706665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,256,0.018760534127553304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,768,0.010729599992434185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,128,0.005719466507434845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,128,0.01917866667111715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,256,0.009221333265304565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,64,0.005562666555245717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,128,0.008687999844551087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,64,0.018428800503412883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,8192,32,0.005543466905752818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,8192,32,0.018461867173512777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,65536,0.3781248092651367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,512,0.009578667084376017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,65536,0.34972801208496096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,16384,0.098908797899882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,16384,0.09087253411610921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,12288,0.07625386714935303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,12288,0.07259519894917807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,12288,0.05712639888127645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,10240,0.06703999837239584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,10240,0.0642848014831543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,16384,0.07520533402760823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,8192,0.05424000024795532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,8192,0.054007466634114584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,10240,0.04925866524378459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,7168,0.047806934515635176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,7168,0.048417067527770995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,8192,0.04123200178146362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,6144,0.04081919987996419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,6144,0.04247466723124187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,7168,0.037034666538238524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,5120,0.03495786587397258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,5120,0.038813865184783934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,6144,0.033701332410176595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,4096,0.028094933430353804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,4096,0.035341866811116535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,5120,0.029150933027267456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,3584,0.02523946762084961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,3584,0.03307519952456157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,4096,0.02521173357963562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,3072,0.021640533208847047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,3072,0.031163734197616578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,3584,0.023130667209625245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,2560,0.018988800048828126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,2560,0.028223999341328937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,3072,0.020626133680343627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,2048,0.01581013302008311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,2048,0.02566933234532674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,2560,0.01820053259531657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,1536,0.012760532895723977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,1536,0.02287999987602234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,2048,0.0161461333433787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,1024,0.009354666868845622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,1536,0.014379733800888061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,1024,0.02120959957440694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,768,0.0077674667040507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,768,0.01954879959424337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,768,0.010776533683141073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,512,0.006185600161552429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,512,0.020083200931549073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,1024,0.011645866433779399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,256,0.0051584000388781226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,256,0.019207467635472618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,512,0.009533866246541341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,128,0.004220800101757049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,256,0.008776533603668212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,128,0.016541866461435954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,64,0.0038783999780813852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,64,0.01734506686528524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,7168,32,0.004138666639725367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,7168,32,0.016499200463294984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,128,0.008760533730189006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,65536,0.33875627517700196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,65536,0.322492790222168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,16384,0.08934186299641927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,16384,0.08627519607543946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,8192,65536,0.2740085283915202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,12288,0.06716159979502359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,12288,0.06850026448567709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,16384,0.044437332948048906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,10240,0.057350401083628336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,10240,0.05970346530278524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,12288,0.03560213247934978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,8192,0.0460863987604777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,8192,0.04984320004781087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,10240,0.030611199140548707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,7168,0.04155733187993367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,7168,0.04342293341954549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,8192,0.026237867275873822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,6144,0.03576320012410482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,6144,0.04053653478622436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,7168,0.024166399240493776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,5120,0.0301418662071228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,5120,0.036534400780995686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,6144,0.022103466590245566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,4096,0.02477226654688517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,4096,0.03342506488164266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,4096,0.01709973414738973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,3584,0.022460800409317017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,3584,0.03125226696332296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,5120,0.019588265816370645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,3072,0.019371734062830607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,3072,0.02945386568705241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,3072,0.014596266547838846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,2560,0.01706026593844096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,2560,0.02706986665725708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,3584,0.01614400049050649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,2048,0.014375467101732889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,2048,0.025406932830810545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,2048,0.012036266922950744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,1536,0.011829333504041036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,1536,0.02321066657702128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,2560,0.01325866679350535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,1024,0.00897279977798462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,1024,0.02087786595026652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,1024,0.00957546631495158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,768,0.0073183998465538025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,768,0.019100799163182577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,1536,0.010410666465759277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,512,0.005865600208441416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,512,0.01854506731033325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,768,0.008788266777992248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,256,0.004646400113900503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,256,0.017100799083709716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,512,0.008388266960779826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,128,0.00421013335386912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,128,0.01632426679134369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,256,0.007877333462238312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,64,0.003815466662247976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,128,0.007499733567237854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,64,0.016826667388280234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,6144,32,0.00386559988061587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,6144,32,0.016528000434239708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,65536,0.2988138516743978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,65536,0.32525014877319336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,16384,0.07917760213216146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,16384,0.08246400356292724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,7168,65536,0.2755039850870768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,12288,0.05944960117340088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,12288,0.06517546574274699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,12288,0.03460906744003296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,10240,0.05038933356602987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,10240,0.05613760153452555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,16384,0.04349546829859416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,8192,0.04203733205795288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,8192,0.04666346708933512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,10240,0.029950932661692305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,7168,0.03625066677729289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,7168,0.04258453448613485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,8192,0.025565866629282636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,6144,0.03210453391075134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,6144,0.03842879931131999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,7168,0.02357333302497864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,5120,0.026973867416381837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,5120,0.035189334551493326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,6144,0.021667200326919555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,4096,0.023107200860977173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,4096,0.0335093339284261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,5120,0.019542400042215982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,3584,0.020132267475128175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,3584,0.031015467643737794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,4096,0.016804265975952148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,3072,0.018718934059143065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,3072,0.02901653250058492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,3584,0.01536853313446045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,2560,0.015610667069753012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,2560,0.026814933617909747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,2560,0.012904533743858337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,2048,0.013300266861915589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,2048,0.024885332584381102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,3072,0.014344533284505209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,1536,0.011256532867749532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,1536,0.02307200034459432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,1536,0.010867200295130412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,1024,0.008462933699289958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,1024,0.02108586629231771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,2048,0.011646933356920878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,768,0.007052800059318543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,768,0.018642133474349974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,6144,65536,0.15815679232279461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,512,0.0059562668204307554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,512,0.018497065703074137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,768,0.008788266777992248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,256,0.004966400067011515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,1024,0.009251200159390767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,256,0.016742400328318276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,128,0.004243200023969015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,128,0.01694186727205912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,512,0.008402132987976074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,128,0.007494399944941203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,64,0.004198400179545084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,64,0.016505600015322367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,5120,32,0.005063466727733612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,256,0.007915733257929485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,5120,32,0.016530133287111917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,65536,0.23265280723571777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,16384,0.0606879989306132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,65536,0.2875328063964844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,16384,0.06872426668802897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,12288,0.046589867273966475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,12288,0.05691413482030233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,12288,0.034061865011850996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,10240,0.039163732528686525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,10240,0.04838826656341553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,16384,0.042284798622131345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,8192,0.032614399989446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,8192,0.041303467750549314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,10240,0.029306666056315107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,7168,0.028544000784556073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,7168,0.03757013479868571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,8192,0.02526719967524211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,6144,0.024987733364105223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,6144,0.035634132226308186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,7168,0.023137066761652628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,5120,0.02143999934196472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,5120,0.03287039995193482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,6144,0.021167999505996703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,4096,0.018041600783665977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,4096,0.030403200785319013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,5120,0.018599466482798258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,3584,0.01600106656551361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,3584,0.028705066442489623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,4096,0.016405333081881204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,3072,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,3072,0.02677226662635803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,3584,0.014958932995796204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,2560,0.012553600470225015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,2560,0.024711465835571288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,3072,0.014113066593805948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,2048,0.01086293359597524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,2048,0.022894932826360067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,2560,0.01242026686668396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,1536,0.008819199601809184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,1536,0.021450666586558025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,2048,0.011667199929555257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,1024,0.006629333396752675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,1536,0.010292266805966694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,1024,0.020115200678507486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,768,0.005885866781075796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,768,0.018846933046976724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,768,0.008733866612116496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,512,0.005026133358478546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,512,0.018729599316914876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,1024,0.008823466300964356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,256,0.004204800228277842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,256,0.016511999567349753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,256,0.00790826678276062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,512,0.008302933474381765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,128,0.0038453333079814913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,128,0.015838932991027833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,64,0.0038005332152048744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,128,0.007501866420110066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,64,0.014816000064214071
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,4096,32,0.0037845333417256674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,4096,32,0.015084800124168397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,65536,0.1974293390909831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,65536,0.2666261355082194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,16384,0.05306666692097982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,16384,0.06503039995829264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,5120,65536,0.1509237289428711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,12288,0.03961600065231323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,12288,0.05156799952189127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,16384,0.04215466578801473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,10240,0.03381866614023844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,10240,0.04482346773147583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,12288,0.033810134728749594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,8192,0.027342933416366576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,8192,0.03903893232345581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,10240,0.029318400224049884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,7168,0.024538666009902954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,7168,0.036662399768829346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,8192,0.02518293261528015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,6144,0.021434666713078816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,6144,0.03321920037269592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,7168,0.02308053374290466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,5120,0.018567466735839845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,5120,0.031137067079544067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,6144,0.021167999505996703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,4096,0.015441067020098367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,4096,0.028834132353464763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,5120,0.018613332509994508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,3584,0.013949867089589438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,4096,0.016155733664830526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,3584,0.02683839996655782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,3072,0.012397866447766621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,3072,0.02609813412030538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,3072,0.014123732844988504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,2560,0.011014399925867717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,2560,0.024114133914311726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,3584,0.015264000495274863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,2048,0.00953493316968282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,2048,0.022987733284632363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,4096,65536,0.14589759508768718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,1536,0.007167999943097432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,1536,0.02169813315073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,1536,0.0100000003973643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,1024,0.005798399945100148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,2560,0.012957866986592612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,1024,0.019245866934458414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,2048,0.0116565336783727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,768,0.005070933202902476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,768,0.02071146567662557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,512,0.004580266773700714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,512,0.017273600896199545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,512,0.008414933085441589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,768,0.00836906631787618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,256,0.0038399999340375268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,256,0.01662720044453939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,128,0.003386666625738144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,128,0.014465066790580749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,1024,0.00922986666361491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,128,0.007480533421039581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,64,0.0031360000371932983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3584,32,0.0034783999125162757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,256,0.007819733520348867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,64,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3584,32,0.016155733664830526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,65536,0.1778048038482666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,16384,0.048596266905466715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,65536,0.24463040033976235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,16384,0.06282026767730713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,12288,0.03712853193283081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,12288,0.04753066698710124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,12288,0.033666133880615234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,10240,0.031033599376678468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,10240,0.042157868544260665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,16384,0.041275731722513836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,8192,0.024665600061416625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,8192,0.03742400010426839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,10240,0.029433600107828778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,7168,0.022536534070968627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,7168,0.034968535105387374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,8192,0.025260800123214723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,6144,0.01898026665051778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,6144,0.031223465998967487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,7168,0.023172267278035483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,5120,0.017116800944010416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,5120,0.029934932788213093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,6144,0.020487467447916664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,4096,0.013917866349220275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,4096,0.027980800469716387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,5120,0.01869759956995646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,3584,0.01286079982916514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,3584,0.026841600735982258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,4096,0.015994667013486227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,3072,0.011076266566912334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,3072,0.024810665845870973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,3584,0.015025066335995993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,2560,0.010171733299891154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,2560,0.023197867472966514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,3072,0.013708800077438354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3584,65536,0.14524906476338703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,2048,0.008569600184758504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,2048,0.023165865739186605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,1536,0.006934399902820587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,1536,0.020692267020543418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,2560,0.012602667013804117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,1024,0.005487999816735586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,1024,0.018959999084472656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,2048,0.011711999773979187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,768,0.004702933132648468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,768,0.017180800437927246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,1536,0.010007466872533162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,1024,0.009180800120035807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,512,0.0038005332152048744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,512,0.017548799514770508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,256,0.003502933432658514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,256,0.016225066781044007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,768,0.008425600330034892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,128,0.0030720000465710956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,128,0.014586666226387024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,256,0.007995733122030894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,64,0.0030773334205150605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,512,0.008307200173536937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,3072,32,0.0030080000559488933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,64,0.01458346645037333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,3072,32,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,128,0.0075530668099721265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,65536,0.15604480107625324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,16384,0.04195626576741536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,65536,0.2318965276082357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,16384,0.05776319901148478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,12288,0.032502400875091556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,12288,0.045076266924540205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,12288,0.03349973360697429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,10240,0.027409066756566364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,10240,0.04075093269348144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,16384,0.04117759863535563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,8192,0.022273067633310953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,8192,0.03556266625722249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,10240,0.029266132911046343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,7168,0.02056106726328532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,7168,0.033469867706298825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,8192,0.02440213362375895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,6144,0.01730666756629944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,6144,0.030953600009282427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,7168,0.0231989324092865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,5120,0.014980266491572062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,5120,0.028938666979471846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,6144,0.020205867290496827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,4096,0.012513066331545511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,4096,0.026948267221450807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,5120,0.018627200524012247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,3584,0.011690666278203327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,3584,0.02592960000038147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,4096,0.016100266575813295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,3072,0.010055466492970785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,3072,0.024602667490641276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,3584,0.014966400464375815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,2560,0.009220266342163086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,2560,0.023086933294932048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,3072,0.013620266318321228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,2048,0.007480533421039581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,2048,0.02182719906171163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,2560,0.012918399771054587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,1536,0.006788266698519389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,1536,0.0209279994169871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,2048,0.01120746632417043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,1024,0.0054624001185099285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,1024,0.01858133276303609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,1024,0.009092266360918682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,768,0.004801066716512045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,768,0.017901867628097534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,1536,0.010471466183662414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,512,0.004284800092379252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,512,0.01678933302561442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,768,0.008750933408737182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,256,0.0034261333445707956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,256,0.01636373301347097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,512,0.008590933680534363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,128,0.003472000112136205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,128,0.01462399959564209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,256,0.007878399888674418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,64,0.003070933371782303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,64,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,128,0.0075647999842961625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2560,32,0.002980266759792964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,65536,0.13941973050435383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2560,32,0.014904533823331198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,65536,0.213865598042806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,16384,0.03848106861114502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,16384,0.05343360106150309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,3072,65536,0.14411412874857585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,12288,0.02962133288383484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,12288,0.042751999696095784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,12288,0.033079467217127484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,10240,0.02451733350753784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,10240,0.03921600182851155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,16384,0.041124268372853594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,8192,0.01989226738611857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,8192,0.03461013237635295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,10240,0.028933332363764448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,7168,0.017901867628097534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,7168,0.03255893389383952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,8192,0.02435200015703837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,6144,0.015449600418408713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,6144,0.02890346646308899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,7168,0.022430932521820067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,5120,0.013874133427937826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,5120,0.027365332841873168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,6144,0.020243199666341145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,4096,0.011410133043924967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,4096,0.024810665845870973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,5120,0.018634666999181114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,3584,0.010436266660690308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,3584,0.026294400294621784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,4096,0.01615466674168905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,3072,0.009001599748929341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,3072,0.022957867383956908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,3584,0.014973866939544677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,2560,0.008039466540018718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,2560,0.02285760045051575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,3072,0.013752532998720804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,2048,0.006673066814740498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,2048,0.0213045338789622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,2560,0.012553600470225015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,1536,0.00598826656738917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,1536,0.021257599194844566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,2048,0.011237333218256634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,1024,0.005081599950790406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,1024,0.018569600582122803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,1536,0.010542933146158855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,768,0.004677333434422811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,1024,0.009156266848246258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,768,0.017058134078979492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,512,0.003966933240493139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,512,0.017205333709716795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2560,65536,0.14310612678527831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,256,0.0035349334279696143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,256,0.015247999628384908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,768,0.008323200047016144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,128,0.0031178665657838186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,512,0.008390399813652038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,256,0.007997866471608479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,128,0.016054399808247886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,64,0.0030975999931494398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,64,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,2048,32,0.003124266614516576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,128,0.007566933333873749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,2048,32,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,65536,0.12108799616495769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,16384,0.033137067159016924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,65536,0.20115200678507486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,16384,0.04943679968516032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,12288,0.024570665756861367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,12288,0.04085013469060262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,12288,0.0325984001159668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,10240,0.020782933632532755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,10240,0.03704426685969035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,16384,0.04113813241322835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,8192,0.01774719953536987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,8192,0.033054933945337935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,10240,0.02879253427187602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,7168,0.015314132968584696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,7168,0.03063360055287679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,8192,0.024322134256362916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,6144,0.013540266950925191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,6144,0.028412799040476482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,7168,0.022275199492772423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,5120,0.011947733163833619
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,5120,0.025960532824198405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,6144,0.02065920035044352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,4096,0.009962667028109233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,4096,0.02515733242034912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,5120,0.018576000134150186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,3584,0.008830933769543966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,3584,0.02456106742223104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,4096,0.016195199886957803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,3072,0.007830399771531422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,3072,0.023545600970586143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,3584,0.014910933375358582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,2560,0.006773333251476288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,2560,0.021576533714930214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,3072,0.013671466708183288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,2048,0.005936000247796377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,2048,0.020866133769353232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,2560,0.012526933352152506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,1536,0.005243733525276184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,1536,0.019316265980402626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,2048,0.011633066336313884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,1024,0.004211199780305227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,1024,0.018228266636530557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,1536,0.009959466258684794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,768,0.0038122666378815973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,768,0.016671999295552572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,768,0.008738133311271667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,512,0.003421866645415624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,1024,0.009152000149091084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,512,0.017435733477274576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,256,0.0029824001093705496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,256,0.014648532867431641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,512,0.008425600330034892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,256,0.007910400132338206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,128,0.0026346666117509207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,128,0.01598186691602071
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,128,0.007594666878382365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,64,0.002603733291228612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,64,0.015558399756749473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1536,32,0.0026506667335828146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1536,32,0.014514133334159851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,65536,0.09661333560943604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,65536,0.18296853701273602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,2048,65536,0.14264639218648273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,16384,0.030114134152730305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,16384,0.051127465565999355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,12288,0.02119893431663513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,12288,0.03926506837209066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,12288,0.032551467418670654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,10240,0.01869973341623942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,10240,0.03625066677729289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,16384,0.0408021330833435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,8192,0.01541866660118103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,8192,0.03297066688537598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,10240,0.02845120032628377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,7168,0.013649066289265951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,7168,0.02831573287645976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,8192,0.024279467264811196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,6144,0.012049067020416259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,6144,0.026833067337671917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,7168,0.022297600905100502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,5120,0.010605866710344952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,5120,0.024915200471878052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,6144,0.02019946575164795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,4096,0.007551999886830647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,4096,0.024699733654658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,5120,0.018220800161361694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,3584,0.006932266553243001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,3584,0.0228383998076121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,4096,0.01614720026652018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,3072,0.006356266637643178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,3072,0.022716800371805825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,3584,0.014892799655596414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,2560,0.005671466886997223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,2560,0.020603734254837035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,3072,0.014110933740933737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,2048,0.005060266455014547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,2048,0.02065599958101908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,2560,0.01250986655553182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,1536,0.004462933540344239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,1536,0.018613332509994508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,2048,0.011492266257603963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,1024,0.0037503999968369803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,1024,0.01890453298886617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,1536,0.010362666845321656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,768,0.0034400001168251038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,768,0.017067732413609822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,1024,0.009205333391825358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,512,0.003047466774781545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,512,0.0160970667997996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,768,0.008717866738637288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,256,0.0025770666698614756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,256,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,512,0.008340266346931458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,256,0.007901866734027863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,128,0.0029397333661715193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,128,0.014664533734321594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,64,0.0025898667673269907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,128,0.007550933460394542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,64,0.016536532839139303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,1024,32,0.0026709333062171934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,1024,32,0.014436266819636025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,65536,0.05934400161107382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,65536,0.17171626091003417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,16384,0.018154666821161906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,16384,0.04491839806238811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1536,65536,0.14285119374593097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,12288,0.016562133034070333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,12288,0.036756265163421634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,12288,0.03236906727155049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,10240,0.014568533500035605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,10240,0.03292160034179688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,16384,0.040994131565093996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,8192,0.01256106694539388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,8192,0.029704533020655316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,10240,0.02805546720822652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,7168,0.01100373367468516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,7168,0.028166399399439497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,8192,0.023960532744725545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,6144,0.009890133142471313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,6144,0.027110399802525838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,7168,0.022345600525538127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,5120,0.008257066706816356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,5120,0.024822400013605753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,6144,0.020423465967178346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,4096,0.007206400235493978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,4096,0.02403200070063273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,5120,0.01816426714261373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,3584,0.006649599969387054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,3584,0.022975999116897582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,1024,65536,0.14358399709065756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,4096,0.01576959987481435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,3072,0.0061034664511680605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,3072,0.023297067483266196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,3584,0.014871467153231302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,2560,0.005435733497142792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,2560,0.02063039938608805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,3072,0.013778133193651834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,2048,0.004774400095144907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,2048,0.020657066504160562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,2560,0.012435199817021687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,1536,0.004265599946180979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,1536,0.018782933553059898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,2048,0.011664000153541566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,1024,0.003764266769091288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,1024,0.018437333901723228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,768,0.0030613332986831666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,768,0.016547200083732606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,1536,0.010186666250228881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,512,0.002979200085004171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,1024,0.0092330664396286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,512,0.015154133240381876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,256,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,768,0.008384000261624653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,256,0.014778666694959006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,128,0.0025087999800841015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,256,0.007974400122960409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,512,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,128,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,64,0.002279466638962428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,768,32,0.002259200066328049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,64,0.014274133245150247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,128,0.007525333265463512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,768,32,0.014991999665896098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,65536,0.04385493199030559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,16384,0.01588373382886251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,65536,0.1646517276763916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,16384,0.04173546632130941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,12288,0.013293866316477457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,12288,0.037170132001241044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,12288,0.03253866632779439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,10240,0.012133333086967468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,10240,0.03304213285446167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,16384,0.04061973492304484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,8192,0.011905066172281901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,8192,0.02935360074043274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,10240,0.02839360038439433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,7168,0.010829866925875346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,7168,0.027340799570083618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,8192,0.024410667022069295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,6144,0.009201066692670186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,6144,0.026756266752878826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,7168,0.022218666474024453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,5120,0.007898666461308797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,5120,0.024742400646209715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,6144,0.020308266083399452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,4096,0.006796800096829732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,4096,0.02393280069033305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,5120,0.018553600708643595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,3584,0.006300800045331319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,3584,0.02268799940745036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,4096,0.016316800316174825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,3072,0.0060917332768440245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,3072,0.02193280061086019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,3584,0.014876799782117209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,2560,0.005475200215975444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,2560,0.02076693375905355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,768,65536,0.14341972668965658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,2048,0.004884266853332519
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,2048,0.02081386645634969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,3072,0.01425386667251587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,1536,0.004315733412901561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,2560,0.012401066223780314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,1536,0.018630399306615194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,1024,0.003522133330504099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,1024,0.017820799350738527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,2048,0.011317333579063416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,1536,0.010360532999038696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,768,0.003437866767247518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,768,0.016922666629155477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,512,0.0029919999341169994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,1024,0.009221333265304565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,512,0.01583253343900045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,256,0.002585600068171819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,768,0.00874133308728536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,256,0.01469013293584188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,128,0.0025631998976071674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,512,0.008413867155710856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,128,0.016689066092173258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,256,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,64,0.0021920000513394673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,128,0.007582933207352956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,512,32,0.002571733295917511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,64,0.01456000010172526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,512,32,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,65536,0.03210666577021281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,16384,0.011266133189201355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,65536,0.15758934020996093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,16384,0.0389301339785258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,12288,0.009544533491134644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,12288,0.033061333497365314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,12288,0.03234986662864685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,16384,0.04047466516494751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,10240,0.008867200215657551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,10240,0.03328746755917867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,10240,0.02844480077425639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,8192,0.008294400076071422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,8192,0.02969493269920349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,7168,0.007613866527875264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,7168,0.02848106622695923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,8192,0.023924267292022704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,6144,0.007106133302052816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,7168,0.02230613430341085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,6144,0.025579732656478883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,5120,0.007929599781831106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,5120,0.024718934297561647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,6144,0.019942400852839152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,4096,0.006807466844717662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,5120,0.018201599518458046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,4096,0.024029866854349772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,3584,0.0063391998410224915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,3584,0.022881066799163817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,512,65536,0.1435413360595703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,3072,0.005635199944178263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,4096,0.016035200158754984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,3072,0.021967999140421548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,2560,0.005212800204753875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,3584,0.014946132898330688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,2560,0.020708266894022623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,2048,0.004705066482226053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,3072,0.013579733173052468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,2048,0.020219733317693077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,1536,0.003975466638803482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,2560,0.012472533186276754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,1536,0.01874986688296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,1024,0.003519999980926514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,2048,0.011166933178901672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,1024,0.016778665781021117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,768,0.003139200061559677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,1536,0.010460799932479859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,768,0.016808533668518068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,1024,0.009123200178146362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,512,0.0027424000203609467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,512,0.014664533734321594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,768,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,256,0.0026229334374268847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,512,0.008135466774304708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,256,0.015702399611473083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,128,0.002288000037272771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,256,0.007559466858704884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,128,0.01490239997704824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,64,0.0022026665508747103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,128,0.00713919997215271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,256,32,0.0022133332987626392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,64,0.014564266800880432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,65536,0.024856533606847128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,256,32,0.014108799894650779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,16384,0.010422399640083313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,65536,0.1535029411315918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,16384,0.03711466789245606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,12288,0.00886826713879903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,12288,0.03303999900817871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,256,65536,0.1443072001139323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,10240,0.00848746697107951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,10240,0.032551467418670654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,16384,0.04067519903182983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,12288,0.03213546673456828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,8192,0.00788266658782959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,8192,0.02938026587168376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,7168,0.007097599903742473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,7168,0.02813653349876404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,10240,0.028431999683380126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,6144,0.0067893331249554946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,6144,0.02558293342590332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,8192,0.023934932549794515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,5120,0.0070592001080513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,7168,0.021887999773025513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,5120,0.024987733364105223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,4096,0.006414933502674103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,4096,0.023496532440185548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,6144,0.02022613286972046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,3584,0.005955199897289276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,3584,0.022692267100016275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,5120,0.018194133043289186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,3072,0.0054666668176651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,4096,0.015777066349983215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,3072,0.02181653380393982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,2560,0.005041066805521647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,3584,0.014890666802724203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,2560,0.020514132579167683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,2048,0.004282666742801667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,3072,0.013736533125241599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,2048,0.020117332537968956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,1536,0.003841066608826319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,2560,0.01246399978796641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,1536,0.019314134120941163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,1024,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,2048,0.011224533120791119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,1024,0.016609066724777223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,1536,0.010499200224876404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,768,0.0029919999341169994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,1024,0.008840533097585042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,768,0.016711467504501344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,512,0.002644266684850057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,768,0.008423466483751934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,512,0.01597866714000702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,256,0.002600533266862233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,512,0.007950933277606964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,256,0.016307199994723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,128,0.00222080002228419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,256,0.00751146674156189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,128,0.014519466956456503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,64,0.002271999915440877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,128,0.007145600020885467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,128,32,0.00223786657055219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,64,0.014484266440073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,128,32,0.014121599992116294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,65536,0.02196586728096008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,16384,0.007901866734027863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,16384,0.03623893260955811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,65536,0.15122772852579752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,12288,0.007112533350785573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,10240,0.006739200154940288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,12288,0.032959999640782674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,768,128,65536,0.14334294001261394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,8192,0.00595306654771169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,10240,0.032561065753300984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,7168,0.005899733304977417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,8192,0.0290175994237264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,6144,0.00619946668545405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,5120,0.006733866532643636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,7168,0.027897600332895917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,6144,0.026774400472640993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,4096,0.0062943999965985615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,3584,0.005906133353710175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,5120,0.025574400027592974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,3072,0.00552106648683548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,4096,0.022823466857274374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,2560,0.0054175997773806255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,3584,0.022806400060653688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,2048,0.004212266703446706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,3072,0.022476800282796226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,2560,0.02071573336919149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,1536,0.003832533210515976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,1024,0.0030293333033720653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,2048,0.019613866011301676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,1536,0.018472532431284584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,768,0.0029728000362714132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,512,0.002621866762638092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,1024,0.016661333044370015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,768,0.016806399822235106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,256,0.002229333420594533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,512,0.015960533420244852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,128,0.0023370665808518727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,256,0.014792533715566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,64,0.0021557333568731946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,64,32,0.002238933245340983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,128,0.014341333508491516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,64,0.014288000265757241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,65536,0.016581333676973977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,64,32,0.013447466492652892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,16384,0.009619200229644775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,12288,0.008699733018875121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,65536,0.15406826337178547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,16384,0.03793386618296306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,10240,0.0074869334697723385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,12288,0.03300159970919291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,8192,0.006667733192443848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,10240,0.03141546646753947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,7168,0.005957333246866862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,8192,0.029205334186553956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,6144,0.006285866598288219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,7168,0.027883734305699664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,6144,0.02712106704711914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,5120,0.006017066538333893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,4096,0.0062730665008227035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,5120,0.02534719904263814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,3584,0.006042666733264923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,3072,0.005523199836413065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,4096,0.02338026762008667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,3584,0.022990934054056802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,2560,0.005497600138187409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,2048,0.004261333247025808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,3072,0.022334933280944824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,1536,0.004268800218900045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,2560,0.02062079906463623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,1024,0.003457066665093104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,2048,0.019723733266194664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,1536,0.019433599710464478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,768,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,1024,0.016806399822235106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,512,0.0026176000634829206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,256,0.0022656001150608064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,768,0.016471466422080992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,512,0.01467519998550415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,128,0.002199466774861018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,256,0.014578133821487427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,64,0.0021664001047611236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,768,32,32,0.0022304000953833262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,128,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,64,0.015184000134468079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,768,32,32,0.013264000415802002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,16384,0.37056214014689126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,16384,0.6369834899902344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,12288,0.4751221338907878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,12288,0.3976352055867513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,10240,0.38334614435831704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,10240,0.28036905924479166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,8192,0.30481386184692383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,8192,0.1948597272237142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,8192,0.1248469352722168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,7168,0.2658954620361328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,7168,0.16354986826578777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,10240,0.15240000089009603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,6144,0.23040533065795898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,6144,0.14316800435384114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,12288,0.1792085329691569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,5120,0.19167466163635255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,5120,0.12173226674397786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,6144,0.09834346771240235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,4096,0.15598613421122234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,4096,0.101801602045695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,7168,0.11102826595306396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,3584,0.1378602663675944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,3584,0.09073279698689779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,5120,0.08436906337738037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,3072,0.11771306991577149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,3072,0.0809109369913737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,16384,0.2371605396270752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,2560,0.10185386339823406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,2560,0.07222719987233481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,3584,0.06409813165664673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,2048,0.0818613370259603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,2048,0.06153066555658976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,4096,0.07054293155670166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,1536,0.06299946705500284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,1536,0.0519157330195109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,3072,0.05680533250172933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,1024,0.04440960089365641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,1024,0.041516800721486405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,2560,0.05035839875539144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,768,0.036183468500773114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,768,0.036525865395863846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,2048,0.04362560113271077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,512,0.027702399094899494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,512,0.031220267216364544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,768,0.02532586654027303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,256,0.018863999843597413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,1024,0.02921066681543986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,256,0.02896639903386434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,128,0.01671253244082133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,128,0.02691199978192647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,1536,0.03683626651763916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,64,0.013447466492652892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,64,0.02678080002466838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,65536,32,0.013896532853444419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,65536,32,0.02754986683527629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,65536,0.6194314956665039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,65536,0.4188224156697591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,128,0.017108267545700072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,16384,0.16337919235229492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,16384,0.11768213113149006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,512,0.02267199953397115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,12288,0.12216533025105794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,12288,0.0934666633605957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,65536,256,0.018669867515563966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,10240,0.10224533081054688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,10240,0.08027626673380533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,10240,0.05090986490249634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,8192,0.08348693052927653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,8192,0.06574933528900147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,12288,0.058754134178161624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,7168,0.07370453675587972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,7168,0.061513598759969076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,16384,0.07479360103607177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,6144,0.06447253227233887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,6144,0.05676906506220499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,8192,0.041767466068267825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,5120,0.05400000015894572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,5120,0.05024960041046143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,7168,0.03794026772181193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,4096,0.0447711984316508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,4096,0.043356800079345705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,5120,0.02996480067571004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,3584,0.03880746761957805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,3584,0.03911999861399333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,4096,0.02532800038655599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,3072,0.0347818652788798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,3072,0.035796264807383224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,6144,0.03405439853668213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,2560,0.029721599817276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,2560,0.032868266105651855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,2560,0.019167999426523842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,2048,0.024612265825271606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,2048,0.029209599892298384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,3584,0.022909865776697794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,1536,0.018997333447138467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,1536,0.025382399559020996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,3072,0.02086720069249471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,1024,0.01420906682809194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,1024,0.02336960037549337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,1024,0.011655466755231221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,768,0.011994666854540507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,768,0.0211136003335317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,1536,0.014526933431625366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,512,0.009378133217493693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,512,0.018732800086339315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,2048,0.016987733046213784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,256,0.007259733478228251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,256,0.020782933632532755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,512,0.009949866930643718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,128,0.00584853341182073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,128,0.01857173244158427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,768,0.010870400071144103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,64,0.006367999811967213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,64,0.018651733795801796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,16384,32,0.006356266637643178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,16384,32,0.019441066185633342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,65536,0.5039818763732911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,128,0.008742400010426839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,65536,0.35152854919433596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,256,0.009609599908192951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,16384,0.12208960056304932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,16384,0.0964949369430542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,12288,0.09375147024790445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,12288,0.07836373647054037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,12288,0.05751146475474039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,10240,0.08039039770762126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,10240,0.06539733409881592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,16384,0.07505173683166504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,8192,0.06441066662470499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,8192,0.05389866828918457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,10240,0.0498741348584493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,7168,0.05724586645762125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,7168,0.052179201443990075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,8192,0.041048534711201984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,6144,0.04873600006103516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,6144,0.048427732785542806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,7168,0.03703573147455851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,5120,0.041818666458129886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,5120,0.04131306807200114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,6144,0.03279680013656616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,4096,0.033088000615437825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,4096,0.037229867776234944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,5120,0.02911253372828166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,3584,0.02974933385848999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,3584,0.03302826682726542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,4096,0.024553600947062174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,3072,0.02568746606508891
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,3072,0.0310698668162028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,3584,0.022634667158126832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,2560,0.02222080032030741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,2560,0.02884693344434102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,3072,0.020523732900619505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,2048,0.01843199928601583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,2048,0.026881066958109538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,2560,0.018706132968266807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,1536,0.014824533462524414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,1536,0.024540799856185912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,1536,0.014077867070833841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,1024,0.011136000355084736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,1024,0.02062293291091919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,2048,0.01611840029557546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,768,0.009212799866994222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,768,0.0197653333346049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,1024,0.011638399958610535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,512,0.0073525334397951765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,512,0.018573866287867228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,768,0.010773332913716634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,256,0.006196266909440359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,256,0.018286933501561485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,256,0.00906773308912913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,128,0.0057781333724657696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,128,0.016546133160591125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,128,0.008754133184750875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,64,0.0054400001962979635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,64,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,12288,32,0.005446400245030721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,12288,32,0.018204800287882485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,512,0.00955733358860016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,65536,0.4218474706013997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,65536,0.31309334437052405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,16384,0.11160639921824138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,16384,0.09623146851857503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,16384,0.0740021308263143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,12288,0.08498240311940511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,12288,0.07594453493754069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,12288,0.05703573226928711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,10240,0.06977813243865967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,10240,0.06005546649297079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,16384,65536,0.2700437227884928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,8192,0.055828265349070226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,8192,0.05013013283411662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,10240,0.04885546763737996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,7168,0.049060265223185226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,7168,0.04816426833470662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,8192,0.040956799189249674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,6144,0.042003198464711504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,6144,0.04417813221613566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,12288,65536,0.2658549308776855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,5120,0.03511146704355876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,5120,0.03879573345184326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,7168,0.03642880121866862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,4096,0.029789867003758748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,4096,0.033317333459854125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,5120,0.02868266701698303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,3584,0.02549546758333842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,3584,0.03115306695302327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,6144,0.033056000868479415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,3072,0.022881066799163817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,3072,0.02993280092875163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,4096,0.024259199698766075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,2560,0.019485867023468016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,2560,0.02845653295516968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,3072,0.020326399803161622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,2560,0.01843199928601583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,2048,0.0160671999057134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,2048,0.025285333395004272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,1536,0.012897066275278726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,1536,0.022768000761667885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,3584,0.022550400098164877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,1024,0.009723732868830364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,1024,0.02056000034014384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,1024,0.011623467008272808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,768,0.00813973347345988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,768,0.018641066551208497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,1536,0.013877333203951518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,512,0.00687360018491745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,512,0.0186901330947876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,768,0.010386133193969726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,256,0.005456000069777171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,256,0.01657919983069102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,512,0.009655466675758362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,128,0.004739200075467428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,256,0.008740267157554627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,128,0.016645333170890807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,64,0.004341333111127218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,64,0.01679253379503886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,10240,32,0.004682666560014089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,10240,32,0.01712533235549927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,128,0.008348799745241801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,65536,0.323088010152181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,65536,0.25523626009623207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,2048,0.016385066509246825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,16384,0.08233599662780762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,16384,0.07457919915517172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,12288,0.06379093329111735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,12288,0.06155840158462524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,12288,0.03381226857503255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,10240,0.05323413213094076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,10240,0.05328960021336874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,16384,0.04210986693700154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,8192,0.04328533411026001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,8192,0.04477333227793376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,10240,0.030282666285832722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,7168,0.03886186679204305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,7168,0.04158399899800618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,8192,0.024825600783030193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,6144,0.03369173208872477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,6144,0.03848426739374797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,7168,0.022961066166559855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,5120,0.028899200757344562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,5120,0.035214932759602864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,10240,65536,0.2724202791849772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,4096,0.023856000105539957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,4096,0.03134613235791524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,6144,0.020459733406702676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,3584,0.021242666244506835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,3584,0.02956053415934245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,5120,0.01874666611353556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,3072,0.01899306575457255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,3072,0.028510934114456175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,4096,0.016587733229001363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,2560,0.01651946703592936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,2560,0.02585279941558838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,3584,0.015148799618085226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,2048,0.013799466689427695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,2048,0.02397973338762919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,3072,0.014327466487884521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,1536,0.011627733707427979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,1536,0.02241706649462382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,2048,0.01134933332602183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,1024,0.008680533369382222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,1024,0.019444266955057778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,2560,0.0125408003727595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,768,0.0073066666722297665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,768,0.018982400496800743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,768,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,512,0.005827199916044871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,512,0.01731626590092977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,1024,0.008724266290664673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,256,0.004621866842110952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,256,0.01649386684099833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,1536,0.010115200281143188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,128,0.004281599819660187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,128,0.017895466089248656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,128,0.007594666878382365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,64,0.003975466638803482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,256,0.00790293316046397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,8192,32,0.005062399804592133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,64,0.01676693360010783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,8192,32,0.01685439944267273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,65536,0.3152501424153646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,512,0.008392533659934998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,65536,0.3294645309448242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,16384,0.07841280301411947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,16384,0.07972266674041747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,12288,0.05834240118662516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,12288,0.0603061318397522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,12288,0.032824534177780154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,10240,0.048478933175404866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,10240,0.05205546617507935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,16384,0.04168533484141032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,8192,0.039876266320546465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,8192,0.04369173447291057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,10240,0.02872213323911031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,7168,0.03440426588058472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,7168,0.04135573307673136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,8192,0.02459733287493388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,6144,0.03078506588935852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,6144,0.03768426577250163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,7168,0.022510933876037597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,5120,0.02550613284111023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,5120,0.03406399885813395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,6144,0.020906666914621987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,4096,0.021195733547210695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,4096,0.030894933144251506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,5120,0.018662399053573607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,3584,0.018594133853912353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,3584,0.02914133270581563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,4096,0.01628266672293345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,3072,0.016821332772572837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,3072,0.02727786699930827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,3584,0.014934399724006652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,2560,0.014233600099881491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,2560,0.025059199333190917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,3072,0.013828266660372415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,2048,0.012469333410263062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,2048,0.02464853326479594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,2560,0.012392533818880717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,1536,0.010048000017801921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,1536,0.020706133047739664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,1536,0.009998933474222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,1024,0.007691733539104462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,1024,0.01874879995981852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,2048,0.011452800035476685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,768,0.006375466783841451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,768,0.01890346606572469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,1024,0.008732799688975017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,512,0.005154133339722951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,512,0.016714666287104288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,768,0.008798933029174805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,256,0.004311466713746389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,256,0.01653333306312561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,256,0.00806933343410492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,128,0.003894400099913279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,128,0.017450666427612303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,512,0.008343467116355896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,64,0.0034517332911491393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,64,0.0163263996442159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,7168,32,0.0034389334420363107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,128,0.00757120003302892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,7168,32,0.015017599860827128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,65536,0.23898666699727378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,65536,0.2284447987874349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,16384,0.06183573404947916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,16384,0.06167680025100708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,8192,65536,0.1472885290781657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,12288,0.04802986780802409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,12288,0.052952531973520914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,16384,0.040744535128275555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,10240,0.03997759819030762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,10240,0.04544639984766642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,12288,0.032849067449569704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,8192,0.03399999936421712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,8192,0.0375434676806132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,10240,0.028509867191314698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,7168,0.02887679934501648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,7168,0.03636266787846883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,8192,0.02414720058441162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,6144,0.026213334004084272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,6144,0.034371201197306314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,6144,0.020248534282048543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,5120,0.022141865889231362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,7168,0.022142932812372843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,5120,0.03146880070368449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,4096,0.0187008003393809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,4096,0.028935466210047407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,4096,0.016238933801651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,3584,0.016914133230845133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,3584,0.028304000695546467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,5120,0.01841813325881958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,3072,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,3072,0.02570880055427551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,3584,0.015130666891733804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,2560,0.013119999567667642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,2560,0.02488320072491964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,3072,0.013085866967837015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,2048,0.011051733295122783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,2048,0.02254293362299601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,2048,0.011130666732788086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,1536,0.009608532985051472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,2560,0.012312533458073933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,1536,0.020984532435735066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,1024,0.007028266787528992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,1024,0.01876373291015625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,7168,65536,0.14113599459330242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,768,0.006009600063165029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,768,0.018731733163197838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,1536,0.010389332969983418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,512,0.005081599950790406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,1024,0.008799999952316284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,512,0.016899200280507405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,256,0.004343466460704803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,256,0.016775466998418174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,768,0.008426666259765625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,128,0.003884800026814143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,128,0.016871466239293417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,128,0.0071946665644645694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,64,0.003853866706291834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,256,0.007959466675917308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,6144,32,0.0038634667793909705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,64,0.015620266397794088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,6144,32,0.01567893326282501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,65536,0.2011669317881266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,65536,0.22070933977762858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,512,0.008321066697438557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,16384,0.05428800185521444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,16384,0.058066133658091226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,12288,0.04092160065968831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,12288,0.04978773196538289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,12288,0.03178559939066569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,10240,0.03483839829762776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,10240,0.04225279887517293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,16384,0.0407968004544576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,8192,0.029397332668304445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,8192,0.03529920180638631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,10240,0.028130133946736652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,7168,0.025251199801762898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,7168,0.034380801518758136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,8192,0.02418346603711446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,6144,0.022456532716751097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,6144,0.03179306586583455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,7168,0.02160853346188863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,5120,0.018709333737691243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,5120,0.031099732716878253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,6144,0.020372267564137778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,4096,0.015838932991027833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,4096,0.027284266551335652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,5120,0.017812265952428182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,3584,0.01397119959195455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,3584,0.026702932516733807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,4096,0.01613759994506836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,3072,0.012787200013796487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,3072,0.024929066499074302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,3584,0.01411626636981964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,2560,0.010909866293271382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,2560,0.02294293244679769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,3072,0.014387200276056925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,2048,0.00992746651172638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,2048,0.020768000682195028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,2560,0.012449066837628682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,1536,0.00790826678276062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,1536,0.019966934124628702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,1536,0.009944533308347065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,1024,0.006045866509278615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,1024,0.018463999032974243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,2048,0.011355732878049214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,768,0.005444266895453135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,768,0.01739413340886434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,1024,0.009106133381525676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,512,0.004598399996757508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,512,0.01889280080795288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,768,0.008843732873598735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,512,0.00795306662718455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,256,0.00393599991997083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,256,0.016394666830698647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,128,0.0034495999415715536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,128,0.014653866489728292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,128,0.0071722666422526045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,64,0.00340693344672521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,64,0.015683199961980185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,5120,32,0.0034133332471052804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,256,0.007779199878374736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,5120,32,0.015403733650843302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,65536,0.1786143938700358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,65536,0.1857024033864339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,16384,0.04749759833017985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,16384,0.05521706740061442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,6144,65536,0.1381696065266927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,12288,0.037360000610351565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,12288,0.045458134015401205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,12288,0.03151999910672505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,10240,0.03144320050875346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,10240,0.03887786865234375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,16384,0.03996800184249878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,8192,0.02527466615041097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,8192,0.03315946658452352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,10240,0.028119466702143353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,7168,0.02283946673075358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,7168,0.033524266878763836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,8192,0.023946666717529298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,6144,0.020037333170572914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,6144,0.03171306649843852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,7168,0.022411733865737915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,5120,0.01678933302561442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,5120,0.029600000381469725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,6144,0.019897600015004478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,4096,0.013900799552599588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,4096,0.0259552001953125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,5120,0.018367999792099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,3584,0.012931199868520102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,3584,0.02537386616071065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,4096,0.01534293293952942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,3072,0.01123199959595998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,3072,0.02298133373260498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,3584,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,2560,0.010225066542625427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,2560,0.02257386644681295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,3072,0.01297813355922699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,2048,0.008551466464996337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,2048,0.020949333906173706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,2560,0.012169599533081055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,1536,0.006790400048096975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,1536,0.02018453280131022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,2048,0.01118293305238088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,1024,0.005006933212280273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,1024,0.017917867501576742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,1536,0.010132267077763876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,768,0.004683733483155568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,768,0.017008000612258913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,1024,0.00870293378829956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,512,0.0038965334494908653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,512,0.017550933361053466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,768,0.00844053328037262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,256,0.0034976000587145484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,256,0.016860800981521606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,512,0.007915733257929485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,256,0.008030933141708375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,128,0.0030271999537944795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,128,0.015314132968584696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,64,0.003110400090614955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,64,0.016402133305867515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,128,0.007470933099587758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,4096,32,0.0031114667654037476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,4096,32,0.015703466534614564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,65536,0.1561087926228841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,65536,0.17912853558858235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,16384,0.04260266621907552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,16384,0.05185706615447998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,5120,65536,0.13536639213562013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,12288,0.03347520033518474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,12288,0.0421674648920695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,16384,0.03965226809183757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,10240,0.02799359957377116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,10240,0.03682773510615031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,12288,0.03138986627260844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,8192,0.023243733247121177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,8192,0.0328874667485555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,8192,0.02364586591720581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,7168,0.019997866948445638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,7168,0.031083732843399048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,10240,0.02775040070215861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,6144,0.017488000790278117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,6144,0.029450666904449464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,7168,0.021545600891113282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,5120,0.01528320014476776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,5120,0.02772586743036906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,6144,0.019509333372116088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,4096,0.012683733304341634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,4096,0.024706133206685386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,4096,0.015299200018246969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,3584,0.011431466539700825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,3584,0.022810665766398112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,5120,0.017442133029301962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,3072,0.010454400380452474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,3072,0.022755199670791627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,3584,0.014051199952761332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,2560,0.009117866555849712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,2560,0.020759467283884683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,3072,0.013346133629480996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,2048,0.007451733450094859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,2048,0.020768000682195028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,4096,65536,0.13431466420491536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,1536,0.005977599819501241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,1536,0.019898666938145956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,2560,0.012086400389671325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,1024,0.004658133288224538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,1024,0.017433599630991618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,2048,0.010822400450706482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,768,0.0042581334710121155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,768,0.017141334215799966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,768,0.008393599589665731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,1024,0.00916266640027364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,512,0.003815466662247976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,512,0.01649066706498464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,256,0.0030016000072161358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,256,0.0156960000594457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,1536,0.010057600339253743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,128,0.0030389333764712016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,128,0.01511573294798533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,256,0.0074890668193499255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,512,0.008416000008583068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,128,0.007601066430409749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,64,0.0030026666820049284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3584,32,0.0030261332790056865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,64,0.016535466909408568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3584,32,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,65536,0.15341013272603352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,65536,0.1570538679758708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,16384,0.04110613266626994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,16384,0.04964160124460856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,12288,0.03166933258374532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,12288,0.04018239974975586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,12288,0.03130666613578796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,10240,0.02648319999376933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,10240,0.03589013417561849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,16384,0.03922453324000041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,8192,0.02160853346188863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,8192,0.031102933486302692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,10240,0.027628799279530842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,7168,0.01908479928970337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,7168,0.0308896005153656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,8192,0.023523199558258056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,6144,0.01760639945665995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,6144,0.02922346591949463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,7168,0.021451733509699502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,5120,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,5120,0.02696853280067444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,6144,0.01951573292414347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,4096,0.012294399738311767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,4096,0.024734934171040855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,5120,0.017361066738764443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,3584,0.011142399907112122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,3584,0.022951465845108033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,4096,0.015561599532763162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,3072,0.010109866658846538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,3072,0.022664533058802287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,3584,0.014064000050226847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,2560,0.008754133184750875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,2560,0.021814399957656862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,3072,0.013318399588267008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,2048,0.007225599884986877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,2048,0.02028053402900696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,2560,0.012006400028864543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,1536,0.006067200005054474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,1536,0.01936639944712321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,1536,0.009546666344006857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,1024,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,1024,0.017838933070500693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,2048,0.010931199789047242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,768,0.00479360024134318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,768,0.016847999890645345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,1024,0.008784000078837078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,512,0.004214400053024292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,512,0.01660160024960836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,768,0.00877333382765452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,256,0.0035274667044480645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,256,0.01474133332570394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3584,65536,0.13388694127400716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,128,0.003457066665093104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,128,0.015471999843915304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,512,0.007922133306662242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,64,0.0029845332105954488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,64,0.014666666587193808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,256,0.007533866663773854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,3072,32,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,3072,32,0.01589120030403137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,128,0.007210666437943776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,65536,0.11950720151265462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,16384,0.03293973406155904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,65536,0.15032000541687013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,16384,0.04498453140258789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,12288,0.025158399343490602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,12288,0.03816959857940674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,12288,0.031269333759943646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,10240,0.02120959957440694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,10240,0.03495786587397258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,16384,0.03912426630655925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,8192,0.018539732694625853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,8192,0.02974613308906555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,10240,0.027215999364852906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,7168,0.015887999534606935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,7168,0.029242666562398274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,8192,0.023229867219924927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,6144,0.014225066701571146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,6144,0.02770773371060689
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,7168,0.02106026609738668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,5120,0.012065066893895467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,5120,0.02641493280728658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,6144,0.019643733898798622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,4096,0.010481066505114238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,4096,0.022825600703557332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,5120,0.017754666010538735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,3584,0.009435733159383137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,3584,0.02353066603342692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,4096,0.015400532881418863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,3072,0.00881813367207845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,3584,0.014046933253606161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,3072,0.022312533855438233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,2560,0.0075765331586201985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,2560,0.020925867557525634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,3072,0.01325653294722239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,2048,0.006552533308664958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,2048,0.01962453325589498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,2560,0.01178666651248932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,1536,0.005547733108202616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,1536,0.01890453298886617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,2048,0.011171199878056844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,1024,0.004598399996757508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,1024,0.01745706597963969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,1536,0.009992532928784688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,768,0.004212266703446706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,768,0.016718933979670204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,1024,0.008738133311271667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,512,0.0038399999340375268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,512,0.01665173371632894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,512,0.007912533481915791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,256,0.003256533294916153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,256,0.01461120049158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,768,0.008391466736793519
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,128,0.0031370667119820913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,128,0.01458133359750112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,256,0.00754559983809789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,64,0.0030026666820049284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,64,0.015094400445620219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,128,0.007272533575693766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2560,32,0.0029994666576385496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,65536,0.09356479644775391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2560,32,0.01523413360118866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,65536,0.13624320030212403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,16384,0.033557331562042235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,16384,0.0434282660484314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,3072,65536,0.13366613388061524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,12288,0.025642667214075727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,12288,0.03764479955037435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,12288,0.03157013257344564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,10240,0.021516799926757812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,10240,0.033701332410176595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,16384,0.03916480143864949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,8192,0.01755626598993937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,8192,0.029480532805124922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,10240,0.027830400069554645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,7168,0.01618666648864746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,7168,0.028036266565322876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,8192,0.023098667462666832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,6144,0.013755733768145243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,6144,0.026708267132441205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,7168,0.021588265895843506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,5120,0.0122079998254776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,5120,0.024809600909550984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,6144,0.019398399194081626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,4096,0.00959999958674113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,4096,0.022632533311843873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,5120,0.017848533391952515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,3584,0.008676266670227051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,3584,0.02287999987602234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,4096,0.015273599823315939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,3072,0.0074314668774604796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,3072,0.02108586629231771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,3584,0.014222932855288186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,2560,0.006562133133411407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,3072,0.01318186620871226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,2560,0.02085226575533549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,2048,0.005965866645177205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,2048,0.02057066758473714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,2560,0.012053333719571431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,1536,0.005083733300367991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,2048,0.010982400178909302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,1536,0.01914026737213135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,1024,0.004242133100827535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,1024,0.017348267634709678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,1024,0.008775466680526733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,768,0.003787733366092046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,768,0.016669867436091106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,1536,0.010082133611043294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,512,0.0036277333895365395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,512,0.015818666418393454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,768,0.008289066453774769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,256,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,256,0.01495680014292399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,512,0.008437333504358928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,128,0.003067733347415924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,256,0.00769706666469574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,128,0.016115199526151022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,64,0.0027242665489514667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,128,0.007066666583220164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,64,0.017384533087412515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,2048,32,0.002976000060637792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,65536,0.081058136622111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,2048,32,0.0150026669104894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,65536,0.12582933108011882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,16384,0.02582719922065735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,16384,0.042761600017547606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2560,65536,0.13357866605122884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,12288,0.018894932667414346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,12288,0.034626134236653644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,12288,0.031325866778691605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,10240,0.016337066888809204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,10240,0.03131626645723979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,16384,0.03870400190353394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,8192,0.013464533289273582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,8192,0.02732586661974589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,10240,0.027080533901850383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,7168,0.012577066818873087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,7168,0.028288000822067262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,8192,0.023089067141215006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,6144,0.010910933216412861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,6144,0.026053333282470705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,7168,0.02122559944788615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,5120,0.009752532839775086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,5120,0.025073067347208662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,6144,0.019010132551193236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,4096,0.009106133381525676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,4096,0.02377386689186096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,5120,0.017459199825922648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,3584,0.008379733562469483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,3584,0.022636799017588298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,4096,0.01532906691233317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,3072,0.007115733126799266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,3072,0.020863999923070274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,3584,0.01421119968096415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,2560,0.006439466774463653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,2560,0.021305600802103676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,3072,0.012871467073758445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,2048,0.005460266768932342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,2048,0.018718934059143065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,2560,0.012677333752314248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,1536,0.004827733337879181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,1536,0.019223467508951823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,2048,0.010808533430099488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,1024,0.004007466634114583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,1024,0.016708266735076905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,1536,0.010055466492970785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,768,0.0035616000493367515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,768,0.016745599110921223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,1024,0.008737066388130188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,512,0.003009066730737686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,512,0.015925332903862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,768,0.008460799853007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,512,0.007914666831493378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,256,0.0030773334205150605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,256,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,128,0.0027456000447273255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,128,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,128,0.007085866729418437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,256,0.007995733122030894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,64,0.0026015999416510267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1536,32,0.002721066772937775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,32,0.01474240024884542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1536,64,0.015010133385658264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,65536,0.05228480100631714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,65536,0.11517013708750408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,16384,0.01722666621208191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,16384,0.039553066094716385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,2048,65536,0.13363626797993977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,12288,0.01707093318303426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,12288,0.033192533254623416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,12288,0.03091199994087219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,10240,0.015046399831771851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,10240,0.029521065950393676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,16384,0.0387605349222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,8192,0.01250879963239034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,8192,0.026816000541051228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,10240,0.027266132831573486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,7168,0.011148800452550251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,7168,0.02542933424313863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,8192,0.023222400744756063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,6144,0.010041600465774536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,6144,0.024932267268498738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,7168,0.021150932709376017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,5120,0.008769067128499348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,5120,0.023217066129048666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,6144,0.019512534141540527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,4096,0.007119999825954437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,4096,0.023450666666030885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,4096,0.015317333738009134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,3584,0.00647680014371872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,5120,0.017352533340454102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,3584,0.022261333465576173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,3072,0.005924266576766968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,3072,0.02084160049756368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,3584,0.014064000050226847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,2560,0.005457066496213277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,3072,0.013236266374588013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,2560,0.01948053240776062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,2048,0.004659200211366018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,2048,0.018875734011332194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,2560,0.012097066640853882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,1536,0.004297600189844767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,2048,0.011212799946467083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,1536,0.01834453344345093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,1024,0.0034901333351929987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,1024,0.016685867309570314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,1024,0.00918293297290802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,1536,0.009971200426419576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,768,0.0033717334270477297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,768,0.01594239970048269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,512,0.0029909332593282064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,512,0.01753066579500834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,768,0.008378666639328004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,256,0.0026581334571043652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,512,0.008044800162315369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,256,0.014784000317255654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,256,0.007604266703128815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,128,0.0026709333062171934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,128,0.014957867066065469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,128,0.007131733496983846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,64,0.0026591998835404714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,64,0.01495146652062734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,1024,32,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,1024,32,0.014966400464375815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,65536,0.04757333199183146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1536,65536,0.1348159948984782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,65536,0.10984960397084553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,16384,0.016376533110936484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,16384,0.03669866720835368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,12288,0.014375467101732889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,12288,0.03388479948043823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,12288,0.03091413378715515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,10240,0.012813867131868998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,10240,0.030938667058944703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,16384,0.03847039937973022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,8192,0.012220799922943115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,8192,0.02657173275947571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,10240,0.02725013295809428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,7168,0.010969600081443787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,8192,0.023101866245269775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,7168,0.026748800277709962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,6144,0.009874133268992107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,6144,0.025043199459711712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,7168,0.021064533789952596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,6144,0.0190175990263621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,5120,0.007895466685295106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,5120,0.024643200635910033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,4096,0.0067445332805315655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,4096,0.02259946664174398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,1024,65536,0.1331210692723592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,3584,0.006331733365853627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,4096,0.015658666690190635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,3584,0.02095359961191813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,5120,0.017442133029301962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,3072,0.005973333120346069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,3072,0.02066133419672648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,2560,0.005532800157864889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,2560,0.0204202671845754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,3072,0.012817066907882691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,2048,0.004700799783070883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,3584,0.014137599865595499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,2560,0.011981866757074992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,1536,0.004211199780305227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,2048,0.018579200903574625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,1536,0.018180267016092936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,1024,0.003832533210515976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,1024,0.016531200210253397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,1536,0.009963732957839967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,768,0.003409066547950109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,1024,0.008794666330019633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,768,0.015898666779200234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,512,0.002979200085004171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,2048,0.010844799876213073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,512,0.01748159925142924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,256,0.0029738667110602063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,256,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,768,0.008381866415341695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,512,0.008398933211962382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,128,0.002679466704527537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,256,0.007515733440717061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,128,0.014702933033307395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,64,0.002566399921973546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,768,32,0.0026015999416510267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,128,0.007207466661930085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,64,0.014672000209490457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,768,32,0.015152000387509666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,65536,0.03211413423220317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,16384,0.012505599856376648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,65536,0.1058026631673177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,16384,0.035027201970418295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,12288,0.011797333757082622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,12288,0.03145706653594971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,12288,0.030555733044942218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,10240,0.010667733351389567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,10240,0.02909226616223653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,16384,0.0387935996055603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,8192,0.011427199840545655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,8192,0.02648746569951375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,10240,0.02683626612027486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,7168,0.009989333152770997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,7168,0.0254528005917867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,8192,0.023179733753204347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,6144,0.009443199634552002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,6144,0.024807467063268026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,7168,0.021127466360727945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,5120,0.007925333579381307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,5120,0.022899200518925987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,6144,0.019451733430226645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,4096,0.006795733173688252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,4096,0.022762666145960488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,5120,0.017378133535385133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,3584,0.006284800171852112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,3584,0.02142613331476847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,4096,0.015357866883277893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,3584,0.014582399527231851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,3072,0.005559466779232025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,3072,0.02081813414891561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,3072,0.012897066275278726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,768,65536,0.13356053034464518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,2560,0.005073066552480062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,2560,0.01948053240776062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,2048,0.004741333425045013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,2048,0.018798933426539103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,1536,0.004023466755946477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,1536,0.018433066209157307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,2560,0.012025599678357441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,1024,0.00342399999499321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,1024,0.016902399063110352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,2048,0.010787199934323628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,768,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,768,0.016263467073440552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,1024,0.00922986666361491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,1536,0.009659733374913533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,512,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,512,0.016124799847602844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,256,0.002625066787004471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,768,0.008297599852085114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,256,0.015850667158762613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,512,0.007885866860548655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,128,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,128,0.014450132846832275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,64,0.002162133405605952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,256,0.007514666517575581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,512,32,0.0021546666820844015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,128,0.007246933380762736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,64,0.01471573313077291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,512,32,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,65536,0.02386346658070882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,16384,0.009595732887585957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,65536,0.09952106475830078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,16384,0.03426880041758219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,12288,0.007996800045172375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,12288,0.031004800399144487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,12288,0.03059200048446655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,10240,0.0100682665904363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,10240,0.02927680015563965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,16384,0.03830506801605225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,8192,0.008871466914812724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,8192,0.027299199501673383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,10240,0.026881066958109538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,7168,0.008382933338483174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,8192,0.02298133373260498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,7168,0.026052266359329224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,6144,0.007667199770609538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,6144,0.024965333938598632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,7168,0.021001599232355752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,5120,0.007149866720040639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,6144,0.019428267081578573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,5120,0.022744532426198324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,4096,0.00664213349421819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,4096,0.020974934101104736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,5120,0.017742933829625447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,3584,0.006250666578610738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,4096,0.01492586632569631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,3584,0.020860799153645835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,3072,0.005710933109124502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,512,65536,0.13437120119730633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,3072,0.02074026664098104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,2560,0.005039999882380167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,2560,0.020361600319544475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,3584,0.014104533195495605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,2048,0.0046517332394917805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,3072,0.012918399771054587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,2048,0.0186901330947876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,1536,0.0038677332301934562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,2560,0.012012799580891926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,1536,0.018497065703074137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,1024,0.003472000112136205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,1024,0.016667733589808144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,1536,0.009970133503278095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,2048,0.010814932982126872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,768,0.0030048000315825146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,768,0.015177599589029946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,512,0.002593066543340683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,1024,0.008784000078837078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,512,0.01695893406867981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,768,0.008355200290679932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,256,0.002571733295917511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,256,0.015291733543078103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,512,0.007551999886830647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,128,0.002229333420594533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,256,0.0075328002373377485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,64,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,128,0.013972266515096029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,128,0.007089066505432129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,256,32,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,64,0.01644266645113627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,65536,0.014498133460680643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,256,32,0.014602667093276978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,16384,0.008804266651471455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,65536,0.09806719621022543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,16384,0.03253866632779439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,12288,0.007579733431339264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,12288,0.029318400224049884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,16384,0.038475732008616134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,256,65536,0.13439040184020995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,10240,0.007326933244864146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,12288,0.030471465984980267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,10240,0.02843093276023865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,8192,0.007203199962774913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,8192,0.026035199562708538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,7168,0.007064533233642578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,10240,0.026808534065882367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,7168,0.025806933641433716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,6144,0.00658133327960968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,8192,0.023095466693242393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,6144,0.025285333395004272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,5120,0.006838400165239971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,5120,0.023779199520746867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,7168,0.02127466599146525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,4096,0.006669866542021434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,4096,0.021614933013916017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,6144,0.019411200284957887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,3584,0.005924266576766968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,3584,0.02122559944788615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,5120,0.017435733477274576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,3072,0.005504000186920166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,4096,0.01529813309510549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,3072,0.020863999923070274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,2560,0.005058133105436961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,3584,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,2560,0.020027732849121092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,65536,0.13326079845428468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,2048,0.004353066782156626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,3072,0.012829867005348206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,1536,0.003952000041802724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,2048,0.01873813271522522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,1536,0.017112533251444496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,2560,0.011966933806737263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,1024,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,2048,0.010803199807802836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,1024,0.01653333306312561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,768,0.002974933385848999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,1536,0.009949866930643718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,768,0.016135467092196147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,512,0.0026015999416510267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,1024,0.008725333213806152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,256,0.002587733417749405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,768,0.008339200417200725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,512,0.014590932925542196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,128,0.002209066599607468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,256,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,512,0.0078005333741505934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,256,0.00763733337322871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,64,0.002204799900452296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,128,0.014546133081118264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,128,32,0.002165333429972331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,512,128,128,0.00717439999183019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,64,0.014707199732462563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,128,32,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,65536,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,16384,0.006293333570162455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,12288,0.006306133170922597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,65536,0.09028800328572592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,16384,0.030834132432937623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,10240,0.005897599955399831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,12288,0.030507733424504597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,8192,0.005870933334032694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,7168,0.005824000140031179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,10240,0.028834132353464763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,8192,0.02486506700515747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,6144,0.005894400179386139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,7168,0.025560534000396727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,5120,0.006346666812896728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,4096,0.005852800110975901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,6144,0.02472213308016459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,5120,0.024938666820526124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,3584,0.005449600021044413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,3072,0.005508266886075338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,4096,0.022680532932281495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,3584,0.020756266514460244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,2560,0.005107200145721436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,3072,0.020629332462946574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,2048,0.005031466484069824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,1536,0.0038304001092910765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,1024,0.0031136001149813337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,2560,0.019006933768590292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,2048,0.019362133741378785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,1536,0.017207467555999757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,512,0.0025770666698614756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,768,0.0029824001093705496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,1024,0.017474132776260375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,768,0.015729066729545594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,256,0.0021664001047611236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,128,0.002268799891074498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,512,0.015220266580581666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,256,0.014589866995811463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,64,0.0021226666867733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,64,32,0.002136533210674922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,128,0.014205867052078247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,64,0.014802133043607077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,64,32,0.014548266927401224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,65536,0.0271722674369812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,16384,0.007098666826883952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,12288,0.007077333331108093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,65536,0.09147946834564209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,16384,0.0310805340607961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,10240,0.006358399987220764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,12288,0.029028266668319702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,8192,0.006289066871007283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,10240,0.02728213270505269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,7168,0.005491200089454651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,8192,0.02658453385035197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,6144,0.005897599955399831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,7168,0.02600319981575012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,5120,0.005524266759554545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,4096,0.005930666625499725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,6144,0.02501973311106364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,3584,0.005897599955399831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,5120,0.02455679972966512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,4096,0.022775467236836752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,3072,0.005592533449331919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,2560,0.005407999952634176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,3584,0.02134933272997538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,3072,0.02093120018641154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,2048,0.005115733544031779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,1536,0.0041696002086003625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,2560,0.02039253314336141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,2048,0.018615466356277467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,1024,0.0031167998909950255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,768,0.0030421334008375804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,1536,0.016696532567342125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,1024,0.017141334215799966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,512,0.002603733291228612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,256,0.002586666742960612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,768,0.016734933853149413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,512,0.016007467110951742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,128,0.0022624000906944276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,64,0.0021418665846188864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,256,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,128,0.014444800217946372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,512,32,32,0.002183466653029124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,64,0.01456106702486674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,512,32,32,0.01458560029665629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,16384,0.285590394337972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,16384,0.49225174585978193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,12288,0.3669568061828613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,12288,0.2230069319407145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,10240,0.31377814610799154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,10240,0.18513919512430826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,8192,0.24766507148742675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,8192,0.1579541365305583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,8192,0.11068267027537029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,7168,0.2159338633219401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,7168,0.13907306989034016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,10240,0.13736106554667155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,6144,0.1920362631479899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,6144,0.129858136177063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,12288,0.16217919985453289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,5120,0.15670933723449706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,5120,0.10082453091939289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,6144,0.08702932993570964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,4096,0.13092693487803142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,4096,0.08689493338267008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,7168,0.09951466719309489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,3584,0.11365653673807781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,3584,0.0760597308476766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,5120,0.07504213651021321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,3072,0.09872106711069742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,3072,0.06811199982961019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,16384,0.2108682632446289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,2560,0.08116052945454916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,2560,0.06044586499532064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,3584,0.05605013370513916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,2048,0.06690239906311035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,2048,0.05290559927622477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,4096,0.06135253508885702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,1536,0.05173866748809815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,1536,0.04488960107167562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,3072,0.049621331691741946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,1024,0.03656853437423706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,1024,0.03704640070597331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,2560,0.04431466658910115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,768,0.02964479923248291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,768,0.0320032000541687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,2048,0.03834026654561361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,512,0.022331732511520385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,512,0.026997333765029906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,768,0.02182719906171163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,256,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,256,0.025000532468159992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,1536,0.03228480021158854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,128,0.012924800316492716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,128,0.0271722674369812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,128,0.014693333705266317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,64,0.013484799861907959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,64,0.02577280004819234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,65536,32,0.014110933740933737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,65536,32,0.02585386633872986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,1024,0.024842667579650878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,65536,0.5013781229654948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,65536,0.32477760314941406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,16384,0.12923946380615234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,16384,0.09895466963450114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,256,0.01611199975013733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,12288,0.10050773620605469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,12288,0.07886719703674316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,65536,512,0.019444266955057778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,10240,0.10127147038777669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,10240,0.07434346675872802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,10240,0.0504800001780192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,8192,0.08301653067270914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,8192,0.06565226713816324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,12288,0.05801173448562622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,7168,0.07224533557891846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,7168,0.05935786565144857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,16384,0.07582933108011881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,6144,0.06338880062103272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,6144,0.05351786613464356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,8192,0.04131199916203816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,5120,0.05477866729100546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,5120,0.04734186728795369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,7168,0.03755840063095093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,4096,0.04516799847284953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,4096,0.04205333391825358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,5120,0.028999465703964233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,3584,0.03896213372548421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,3584,0.0375765323638916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,6144,0.03347520033518474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,3072,0.03342293500900269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,3072,0.033080534140268965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,4096,0.024617600440979003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,2560,0.028939733902613323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,2560,0.03097066680590312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,2560,0.01866453289985657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,2048,0.023345067103703817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,2048,0.027560534079869588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,3072,0.019883733987808228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,1536,0.017937066157658894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,1536,0.024769065777460735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,3584,0.022282665967941283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,1024,0.013033599654833476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,1024,0.021802665789922078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,1024,0.011506133278210958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,768,0.010686933000882467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,768,0.02026026646296183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,1536,0.014127999544143677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,512,0.008351999521255492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,512,0.01866453289985657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,768,0.010475732882817586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,256,0.0059562668204307554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,256,0.01911466717720032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,2048,0.016800000270207723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,128,0.006051200131575266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,128,0.018205867211023966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,256,0.009194667140642803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,512,0.009556266665458679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,64,0.0054954667886098225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,16384,32,0.005895466605822245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,64,0.01665493349234263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,16384,32,0.01814080079396566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,65536,0.3696565310160319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,65536,0.2611029307047526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,128,0.009156266848246258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,16384,0.09897706508636475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,16384,0.0824682633082072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,12288,0.07491626739501953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,12288,0.06445013284683228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,12288,0.03511999845504761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,10240,0.07086933453877767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,10240,0.06137706836064657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,16384,0.044420266151428224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,8192,0.0577834685643514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,8192,0.05486079851786295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,10240,0.030842665831247968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,7168,0.05051519870758057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,7168,0.04996159871419271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,8192,0.02582613428433736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,6144,0.0449621319770813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,6144,0.0452394684155782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,7168,0.02342080076535543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,5120,0.03768213192621867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,5120,0.03985706567764282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,6144,0.02106026609738668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,4096,0.03128640055656433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,4096,0.03376426696777344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,5120,0.018832000096638997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,3584,0.027797333399454754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,3584,0.031215999523798627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,4096,0.01633386711279551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,3072,0.0240064005057017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,3072,0.029564799865086873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,3584,0.015440000096956888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,2560,0.020377600193023683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,2560,0.02680319945017497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,3072,0.014126933614412942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,2048,0.01657386620839437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,2048,0.025176533063252765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,2560,0.013379200299580892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,1536,0.013209600249926248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,1536,0.023113600413004556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,1536,0.010479999581972758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,1024,0.010016000270843506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,1024,0.020037333170572914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,2048,0.011250133315722149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,768,0.00848746697107951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,768,0.019177599747975668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,768,0.008335999647776286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,512,0.006141866743564606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,512,0.01772693395614624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,1024,0.009202133615811665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,256,0.004882133503754934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,256,0.018293333053588868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,512,0.008316799998283386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,128,0.004199466605981191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,128,0.01480959951877594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,256,0.007894399762153625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,64,0.003915733347336451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,64,0.016581333676973977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,12288,32,0.003978666663169861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,12288,32,0.01646933356920878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,128,0.007089066505432129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,65536,0.31284478505452473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,65536,0.2530453364054362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,16384,0.08278186321258545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,16384,0.07902719974517822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,16384,0.04178773164749146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,12288,0.06340159972508749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,12288,0.06080000003178915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,12288,0.0331989328066508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,10240,0.05378133455912272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,10240,0.05795199871063232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,16384,65536,0.2706986745198568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,8192,0.0447221318880717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,8192,0.04752426544825236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,10240,0.02913386623064677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,7168,0.038344534238179524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,7168,0.04325120051701863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,8192,0.024729599555333458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,6144,0.03468586603800456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,6144,0.037828266620635986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,12288,65536,0.1504213333129883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,5120,0.0291103998819987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,5120,0.03354026476542155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,7168,0.02211093306541443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,4096,0.02417280077934265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,4096,0.030885332822799684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,6144,0.02053440014521281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,3584,0.02146240075429281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,3584,0.029546666145324706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,5120,0.018224000930786133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,3072,0.018564265966415406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,3072,0.0267850657304128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,4096,0.015961600343386333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,2560,0.016405333081881204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,2560,0.024951465924580894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,3584,0.015219199657440185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,2048,0.014419200023015341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,2048,0.023861332734425863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,3072,0.013772799571355184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,1536,0.011431466539700825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,1536,0.02138026754061381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,2048,0.011476266384124755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,2560,0.01250986655553182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,1024,0.009161600470542907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,1024,0.02099519968032837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,768,0.007595733304818471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,768,0.019441066185633342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,768,0.008376533786455791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,512,0.005982933441797892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,512,0.018387200435002644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,1536,0.010005333026250203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,256,0.005049600203831991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,256,0.016523733735084534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,1024,0.008923733234405517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,128,0.004305066665013632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,128,0.016717867056528727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,256,0.007528533538182576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,64,0.0038762666285037995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,64,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,10240,32,0.005117866893609365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,10240,32,0.016586666305859886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,128,0.007228800157705943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,65536,0.29324585596720376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,65536,0.21888213157653807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,16384,0.08021439711252848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,512,0.00807360013326009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,16384,0.07229759693145751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,12288,0.06049813429514567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,12288,0.05722560087839762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,12288,0.03208426634470622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,10240,0.05128000179926554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,10240,0.05018986860911051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,16384,0.04041386842727661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,8192,0.04203093449274699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,8192,0.045244801044464114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,10240,0.028333866596221925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,7168,0.03691946665445964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,7168,0.04119999806086223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,8192,0.02357333302497864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,6144,0.03256640036900838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,6144,0.03529813289642334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,7168,0.02191466689109802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,5120,0.027738666534423827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,5120,0.03328533371289571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,6144,0.019871999820073448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,4096,0.022307199239730836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,4096,0.0299615999062856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,10240,65536,0.14214399655659993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,3584,0.020163200298945107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,3584,0.028707200288772584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,5120,0.018243199586868285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,3072,0.017352533340454102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,3072,0.02621440092722575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,3584,0.014988799889882406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,2560,0.014987732966740927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,2560,0.025378133853276568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,4096,0.015754666924476624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,2048,0.012685867150624594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,2048,0.023299199342727662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,3072,0.013347199559211731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,1536,0.010389332969983418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,1536,0.02063039938608805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,2048,0.010774399836858113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,1024,0.007919999957084655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,1024,0.01917653282483419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,2560,0.012476799885431926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,768,0.006691200037797292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,768,0.0185098667939504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,1536,0.010250666737556457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,512,0.005098666747411092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,512,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,768,0.008353066444396973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,256,0.004286933441956838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,256,0.016708266735076905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,512,0.008310399949550629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,128,0.0042463997999827065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,128,0.016452266772588094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,128,0.007162666817506154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,64,0.0038058665891488397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,1024,0.008780800302823384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,256,0.0075541332364082335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,64,0.01653333306312561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,8192,32,0.004029866556326548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,8192,32,0.016528000434239708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,65536,0.19874240557352701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,65536,0.28487253189086914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,16384,0.07465173403422037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,16384,0.0645578662554423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,12288,0.05758506854375204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,12288,0.05530986785888672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,12288,0.031830400228500366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,10240,0.04912213484446208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,10240,0.04549546639124553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,16384,0.03972586790720622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,8192,0.035004798571268717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,8192,0.040635732809702556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,10240,0.028109866380691528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,7168,0.030483200152715044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,7168,0.03700480063756307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,8192,0.02318399945894877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,6144,0.027210666735966997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,6144,0.03307200074195862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,7168,0.02151040037473043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,5120,0.022869332631429037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,5120,0.03093973398208618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,6144,0.019689599672953285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,4096,0.018742400407791137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,4096,0.02725119988123576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,5120,0.018122667074203493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,3584,0.0167797327041626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,3584,0.026825600862503053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,4096,0.015386666854222616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,3072,0.014577066898345948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,3072,0.024885332584381102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,3584,0.014148267110188803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,2560,0.012821333607037863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,2560,0.022844799359639487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,3072,0.01288426617781321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,2048,0.010863999525705974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,2048,0.021121066808700562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,2560,0.012039466698964437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,1536,0.00906773308912913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,1536,0.02003413240114848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,1536,0.009591466188430786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,1024,0.006920533378918965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,1024,0.018389334281285606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,2048,0.010918399691581726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,768,0.0056874667604764305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,768,0.017411200205485027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,1024,0.008988799651463826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,512,0.004705066482226053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,512,0.017398399114608765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,768,0.008714666962623597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,256,0.0040522667268912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,256,0.016060800353686015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,512,0.007890133559703827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,128,0.0034048000971476236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,128,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,256,0.0075445334116617845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,64,0.0033770665526390077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,64,0.014843733112017313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,7168,32,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,128,0.007114666700363159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,7168,32,0.01653866668542226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,65536,0.2272437254587809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,65536,0.1729973316192627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,16384,0.06157013177871704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,16384,0.062156800429026285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,8192,65536,0.13526612917582195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,12288,0.046380798021952316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,12288,0.04870293140411377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,16384,0.03980799913406372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,10240,0.040319999059041337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,10240,0.043900799751281736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,12288,0.03102506597836812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,8192,0.03243946631749471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,8192,0.037390931447347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,10240,0.027461334069569902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,7168,0.029022934039433797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,7168,0.03520533243815104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,8192,0.023563732703526817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,6144,0.024921600023905435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,6144,0.03140160044034322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,6144,0.019463467597961425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,5120,0.020924800634384157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,5120,0.02912000020345052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,7168,0.021670399109522502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,4096,0.01762666702270508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,4096,0.027204267183939618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,5120,0.017429333925247193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,3584,0.015269333124160766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,3584,0.025561600923538208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,4096,0.015082666277885437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,3072,0.01383039951324463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,3072,0.02483839988708496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,3072,0.012990933656692506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,2560,0.012285866340001424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,2560,0.02228053410847982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,3584,0.014082133769989014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,2048,0.010128000378608703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,2048,0.02067199945449829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,2048,0.010469333330790202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,1536,0.008265600105126699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,1536,0.018692266941070557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,2560,0.012157866358757019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,1024,0.0061152001221974695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,1024,0.016874667008717856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,1024,0.009217066566149394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,768,0.005098666747411092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,768,0.018196266889572144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,1536,0.009963732957839967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,512,0.0042357335488001505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,512,0.017691733439763387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,512,0.007885866860548655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,256,0.003453866640726725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,256,0.016199466586112977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,768,0.008498133222262064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,128,0.0032106667757034303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,128,0.014612266421318054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,7168,65536,0.13412373860677082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,256,0.007494399944941203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,128,0.007551999886830647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,64,0.0030847998956839246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,6144,32,0.003091199944416682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,64,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,6144,32,0.015450666348139444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,65536,0.19879147211710613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,65536,0.16250559488932292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,16384,0.05357973178227743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,16384,0.05678720076878866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,12288,0.04051839907964071
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,12288,0.04502933422724406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,12288,0.03094826738039653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,10240,0.03444480101267497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,10240,0.039468801021575926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,16384,0.03931093215942383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,8192,0.02908693353335063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,8192,0.03624746799468994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,10240,0.027322665850321455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,7168,0.024845866362253825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,7168,0.0330570658047994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,8192,0.023486934105555215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,6144,0.022555732727050783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,6144,0.031114667654037476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,7168,0.0214954674243927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,5120,0.01830186645189921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,5120,0.028128000100453694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,6144,0.01983893314997355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,4096,0.015525333086649575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,4096,0.02661866744359334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,5120,0.0173418660958608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,3584,0.013800533612569174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,3584,0.02482773264249166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,4096,0.015146666765213012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,3072,0.012489599982897441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,3072,0.023113600413004556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,3584,0.014084266622861228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,2560,0.010804266730944315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,2560,0.02102186679840088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,3072,0.01292693316936493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,2560,0.011611732840538024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,2048,0.009543466567993163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,2048,0.02039466698964437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,1536,0.007590400179227193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,1536,0.01885546644528707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,2048,0.011025066177050274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,1024,0.006071466704209646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,1536,0.009667199850082398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,1024,0.019223467508951823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,768,0.00525546669960022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,768,0.017303466796875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,6144,65536,0.13167786598205566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,512,0.004357333481311798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,512,0.017783466974894205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,512,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,256,0.0038762666285037995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,256,0.015863466262817382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,768,0.008386133114496867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,128,0.003455999990304311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,128,0.014918399850527444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,128,0.007201066613197327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,1024,0.008740267157554627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,256,0.007527466615041096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,64,0.0030805334448814393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,5120,32,0.0030720000465710956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,64,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,5120,32,0.014732799927393594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,65536,0.1523808002471924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,65536,0.14498987197875976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,16384,0.04158933162689209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,16384,0.05310826698939005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,12288,0.032202666997909545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,12288,0.04162240028381348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,12288,0.030529065926869707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,10240,0.027722666660944622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,10240,0.03701759974161784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,16384,0.038431998093922934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,8192,0.02244159976641337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,8192,0.032562132676442465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,10240,0.026833067337671917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,7168,0.019403733809789023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,7168,0.030308266480763752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,8192,0.02307413419087728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,6144,0.01731946667035421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,6144,0.028652799129486085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,7168,0.0210698664188385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,5120,0.014672000209490457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,5120,0.0267850657304128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,6144,0.019342933098475137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,4096,0.012337066729863485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,4096,0.0244159996509552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,5120,0.01693333387374878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,3584,0.011132799585660299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,3584,0.02285760045051575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,4096,0.014988799889882406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,3072,0.010307199756304423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,3072,0.020793600877126058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,3584,0.01366933286190033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,2560,0.008807466427485148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,2560,0.02027733325958252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,3072,0.01300159990787506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,2048,0.007682133217652638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,2560,0.012029866377512615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,2048,0.0200981338818868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,1536,0.006037333110968272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,1536,0.018685867389043175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,1536,0.009732266267140705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,1024,0.005478399991989136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,1024,0.017028266191482545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,2048,0.01074133316675822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,768,0.0047082667549451195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,768,0.016607999801635742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,1024,0.008989866574605305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,512,0.00425493319829305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,512,0.017178666591644288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,768,0.008302933474381765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,256,0.0034048000971476236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,256,0.01560533344745636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,256,0.007910400132338206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,128,0.0034005333979924522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,128,0.01535040040810903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,512,0.007990399996439617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,64,0.003138133386770884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,64,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,4096,32,0.0031061333914597826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,128,0.0071370666225751235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,4096,32,0.01467519998550415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,65536,0.1604640007019043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,65536,0.1329248030980428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,16384,0.04396373430887858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,16384,0.04970879952112834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,5120,65536,0.13061227003733317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,12288,0.03333119948705037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,12288,0.038702932993570964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,16384,0.03828906615575155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,10240,0.02847893238067627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,10240,0.03559786478678385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,12288,0.03065920074780782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,8192,0.023061333100001018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,8192,0.032580266396204635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,8192,0.023196800549825033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,7168,0.020082134008407592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,7168,0.030743465820948286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,10240,0.027007999022801717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,6144,0.017900800704956053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,6144,0.02890026569366455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,7168,0.021101866165796915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,5120,0.015426133076349893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,5120,0.026714666684468584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,6144,0.019040000438690186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,4096,0.012759466965993246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,4096,0.024651734034220378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,4096,0.014936533570289613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,3584,0.011415466666221619
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,3584,0.02318293253580729
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,5120,0.01693013310432434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,3072,0.010450133681297302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,3072,0.021563732624053956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,3584,0.014052266875902811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,2560,0.009079466263453167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,2560,0.020326399803161622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,3072,0.012872533003489176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,2048,0.0077802668015162155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,2048,0.01909546653429667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,4096,65536,0.12994133631388347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,1536,0.0063274666666984555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,1536,0.01858560045560201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,2048,0.010414933164914448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,1536,0.009597866733868917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,1024,0.00421973317861557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,1024,0.017273600896199545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,768,0.0038581334054470064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,768,0.016849066813786825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,2560,0.012051199873288471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,512,0.003433600068092346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,512,0.015727999806404113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,512,0.008055466910203297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,256,0.002998399982849757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,256,0.01569919983545939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,768,0.008705066641171773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,128,0.003020799905061722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,128,0.014636799693107605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,1024,0.008705066641171773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,64,0.0026506667335828146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,256,0.007501866420110066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,128,0.007516799867153168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3584,32,0.0029994666576385496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,64,0.014792533715566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3584,32,0.014565333724021912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,65536,0.13219733238220216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,65536,0.125383464495341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,16384,0.03575679858525594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,16384,0.04328533411026001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,12288,0.027537065744400024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,12288,0.03505599896113078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,12288,0.030514132976531983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,10240,0.023471999168395995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,10240,0.03309333324432373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,16384,0.03825920025507609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,8192,0.019214934110641478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,8192,0.030451200405756634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,10240,0.026846933364868163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,7168,0.016770132382710776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,7168,0.02895680069923401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,8192,0.022900267442067464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,6144,0.014885333180427552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,6144,0.027009065945943194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,7168,0.02069546580314636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,5120,0.012878933548927307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,5120,0.024707200129826863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,6144,0.019048533837000527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,4096,0.0109333336353302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,4096,0.022679466009140014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,5120,0.016948266824086507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,3584,0.009962667028109233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,3584,0.021258666117986044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,4096,0.014939733346303306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,3072,0.008889599641164144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,3072,0.020718934138615926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,3584,0.01409280002117157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,2560,0.007613866527875264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,2560,0.019102933009465535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,3072,0.012857600053151449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,2048,0.006438399851322174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,2048,0.019146666924158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,2560,0.0120469331741333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,1536,0.005449600021044413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,1536,0.0174453337987264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,2048,0.010833066701889039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,1024,0.004314666489760081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3584,65536,0.12990933259328205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,1536,0.009603200356165568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,1024,0.01655893325805664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,768,0.0038293334345022834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,768,0.015736533204714458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,512,0.0034677334129810332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,512,0.01601066688696543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,512,0.007942399879296621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,256,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,768,0.008307200173536937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,256,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,128,0.0026591998835404714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,128,0.014467199643452963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,1024,0.008721066514650981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,256,0.007961600025494894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,64,0.002630399912595749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,64,0.01483519971370697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,128,0.007539199789365132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,3072,32,0.0026528000831604003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,3072,32,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,65536,0.1093066692352295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,65536,0.11987733046213786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,16384,0.03516906499862671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,16384,0.045772798856099445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,12288,0.027035733064015705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,12288,0.03580586512883504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,12288,0.03063253362973531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,10240,0.022461867332458495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,10240,0.03263573249181111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,16384,0.0381877342859904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,8192,0.017858133713404337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,8192,0.028830933570861816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,10240,0.02730453411738078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,7168,0.016364799936612447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,7168,0.027485867341359455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,8192,0.022312533855438233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,6144,0.01421333352724711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,6144,0.0247488001982371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,7168,0.020758400360743202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,5120,0.012593066692352295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,5120,0.02288960019747416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,6144,0.018634666999181114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,4096,0.009524266918500264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,4096,0.02217493255933126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,5120,0.01700586676597595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,3584,0.008749866485595703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,4096,0.014957867066065469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,3584,0.021604265769322714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,3072,0.00796693315108617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,3072,0.02073813279469808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,3072,0.013026133179664612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,3584,0.013646933436393737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,2560,0.006773333251476288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,2560,0.01925440033276876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,2048,0.006260266900062561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,2048,0.019186133146286012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,2048,0.010444800059000652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,1536,0.005454933146635691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,1536,0.017913599809010826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,2560,0.012110933661460876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,1024,0.004366933306058248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,1024,0.016780799627304076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,1024,0.008824533224105835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,768,0.003942399968703588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,768,0.01671253244082133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,1536,0.009595732887585957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,512,0.0034485332667827605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,512,0.017240534226099648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,512,0.00791786660750707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,256,0.003155199935038885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,256,0.014664533734321594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,768,0.00843946635723114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,128,0.0030730667213598887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,128,0.016362667083740234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,256,0.007480533421039581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,64,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,64,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2560,32,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,128,0.007242666681607564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2560,32,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,65536,0.09868693351745605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,65536,0.1074069341023763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,16384,0.031013333797454835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,16384,0.04230613311131795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,3072,65536,0.13013546466827391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,12288,0.01993066668510437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,12288,0.03237119913101196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,12288,0.030653866132100423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,10240,0.017068799336751303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,10240,0.030418133735656737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,16384,0.03798186779022217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,8192,0.01418773333231608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,8192,0.02850666642189026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,10240,0.026909865935643512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,7168,0.013040000200271606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,7168,0.027024000883102417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,8192,0.02228053410847982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,6144,0.01130560040473938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,6144,0.025018666187922162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,7168,0.0206986665725708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,5120,0.009923199812571209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,5120,0.02371306618054708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,6144,0.018602667252222697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,4096,0.008473599950472515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,4096,0.02241599957148234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,5120,0.017407999436060587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,3584,0.007964799801508587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,4096,0.014929067095120749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,3584,0.021447465817133585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,3072,0.007237333556016285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,3072,0.02069013317426046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,3072,0.013031466801961263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,3584,0.013705600301424661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,2560,0.0062943999965985615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,2560,0.020746666193008422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,2048,0.005688533186912537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,2048,0.019820799430211387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,2560,0.011684266726175944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,1536,0.004828799764315287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,1536,0.018669867515563966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,1536,0.009905067086219788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,2048,0.010870400071144103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,1024,0.00413226659099261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,1024,0.016778665781021117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,768,0.003703466554482778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,768,0.01632533371448517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,768,0.008269866804281871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,512,0.003005866706371307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,512,0.015438933173815408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,1024,0.008821333448092144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,256,0.0036149332920710243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,256,0.015053866306940713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,512,0.007716266810894013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,128,0.0029887999097506206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,256,0.0076330666740735366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,128,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,64,0.0025802666942278544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,64,0.014435199896494546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,2048,32,0.002629333237806956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,128,0.007534933090209961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,2048,32,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2560,65536,0.1301909367243449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,65536,0.07144853274027506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,16384,0.022949333985646567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,65536,0.09606613318125407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,16384,0.0382421334584554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,12288,0.019065600633621217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,12288,0.031092266241709392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,12288,0.030585600932439165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,10240,0.015388799707094827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,10240,0.02906026641527812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,16384,0.03790826797485351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,8192,0.012633599837621055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,8192,0.026796799898147584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,10240,0.026420267422993977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,7168,0.01167039970556895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,7168,0.02683626612027486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,8192,0.022252800067265828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,6144,0.01009493370850881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,6144,0.02421440084775289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,7168,0.020679465929667153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,5120,0.009171199798583985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,5120,0.02309653361638387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,6144,0.01893226703008016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,4096,0.0075765331586201985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,4096,0.021514666080474854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,4096,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,3584,0.006757333377997081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,3584,0.02072319984436035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,5120,0.017080533504486083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,3072,0.006120533247788747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,3072,0.020654932657877604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,3584,0.013699199755986533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,2560,0.005413333574930826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,2560,0.019033600886662803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,3072,0.0130431999762853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,2048,0.004790399968624115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,2048,0.019655466079711914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,2560,0.01183893382549286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,1536,0.004218666752179464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,1536,0.01692906618118286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,2048,0.010856533050537109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,1024,0.0035946667194366455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,1024,0.017172267039616905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,1536,0.00958720048268636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,1024,0.008898133039474487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,768,0.0031199999153614043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,768,0.01552959978580475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,512,0.0030346666773160298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,512,0.015252266327540079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,768,0.008340266346931458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,256,0.002605866640806198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,256,0.014500266313552857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,512,0.007949866851170858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,128,0.002586666742960612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,2048,65536,0.13006400267283122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,128,0.014439466595649719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,256,0.007538133362929027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,64,0.002381866673628489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1536,32,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,64,0.01350826621055603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,128,0.007597866654396057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,65536,0.053381331761678065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1536,32,0.014683733383814493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,16384,0.01709653337796529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,65536,0.09252479871114096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,16384,0.03555200099945068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,12288,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,12288,0.033114665746688844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,12288,0.029736532767613726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,10240,0.013075199723243714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,10240,0.02881493369738261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,16384,0.037588266531626384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,8192,0.012422399719556172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,8192,0.02600533366203308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,10240,0.026029866933822633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,7168,0.011122133334477742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,7168,0.02510400017102559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,8192,0.022741333643595377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,6144,0.009642666578292847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,6144,0.023516800006230673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,7168,0.020660267273585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,5120,0.008359466989835102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,5120,0.022829866409301756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,5120,0.017378133535385133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,4096,0.007118933399518331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,4096,0.02188373406728109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,6144,0.019027199347813925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,3584,0.0064181332786877945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,3584,0.02081706722577413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,3584,0.013722667098045349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,3072,0.006054399907588959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,4096,0.014918399850527444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,3072,0.019517866770426433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,2560,0.005449600021044413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,2560,0.01978879968325297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,2560,0.011623467008272808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,2048,0.004664533336957296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,2048,0.019425066312154134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,3072,0.012852266430854797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,1536,0.004283733169237773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,1536,0.01726400057474772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,1536,0.010041600465774536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,1024,0.003804799914360046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,2048,0.010794666409492493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,1024,0.016670932372411094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,768,0.0034847999612490333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,768,0.016390400131543477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,1024,0.008730666836102803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,768,0.008389332890510559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,512,0.003009066730737686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,512,0.01739306648572286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,256,0.0029557332396507262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,256,0.015429332852363586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,512,0.007683200140794118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,128,0.0026229334374268847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,256,0.00766293356815974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,128,0.015438933173815408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,64,0.00262719988822937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,64,0.014517333110173544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,128,0.007209600011507671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,1024,32,0.0026047999660174055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,65536,0.040777599811553954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,1024,32,0.014455466469128927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,65536,0.08641386826833089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,16384,0.013124266266822815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,16384,0.033852799733479814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1536,65536,0.12975786526997884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,12288,0.010875733693440755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,12288,0.02955840031305949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,12288,0.029745066165924074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,10240,0.011098666985829671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,10240,0.026898133754730224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,16384,0.03736426830291748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,8192,0.01178666651248932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,8192,0.027063467105229694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,10240,0.026372265815734864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,7168,0.010472533106803895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,7168,0.024732800324757893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,8192,0.02232426603635152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,6144,0.008975999553998311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,6144,0.02294293244679769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,7168,0.0206496000289917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,5120,0.007929599781831106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,5120,0.022167466084162393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,6144,0.018982400496800743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,4096,0.006695466736952464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,4096,0.020966400702794395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,1024,65536,0.12980799674987792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,3584,0.006400000055631001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,4096,0.014963199694951376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,3584,0.020885332425435384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,5120,0.01703146696090698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,3072,0.005795200169086456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,3072,0.01975359916687012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,2560,0.005128533144791921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,2560,0.018926932414372762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,2560,0.011666133006413778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,3584,0.013741866747538248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,3072,0.012499200304349263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,2048,0.0047189335028330484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,1536,0.003955200066169103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,2048,0.018657066424687705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,1536,0.016764799753824867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,1024,0.003522133330504099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,1024,0.01649493376413981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,1536,0.009629866480827332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,768,0.003031466652949651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,1024,0.00848426620165507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,2048,0.010476799805959065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,768,0.015715199708938598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,512,0.002587733417749405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,512,0.01627413332462311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,768,0.008541867136955261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,512,0.007910400132338206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,256,0.0026762666801611584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,256,0.01602773368358612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,128,0.002253866692384084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,128,0.014317867159843446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,256,0.007390933235486348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,64,0.002299733211596807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,128,0.007129600147406261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,768,32,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,64,0.013982933759689332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,768,32,0.013756799697875976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,65536,0.030982400973637896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,65536,0.07949546972910562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,16384,0.011689600348472596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,16384,0.030922667185465498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,12288,0.010201600193977357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,12288,0.02786133289337158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,16384,0.03740053176879883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,10240,0.00907306671142578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,10240,0.026781866947809856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,12288,0.030004266897837324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,8192,0.008089600006739299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,8192,0.026104533672332765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,10240,0.02600319981575012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,7168,0.007480533421039581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,7168,0.024658133586247764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,8192,0.02240426739056905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,6144,0.007214933137098948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,6144,0.02363626758257548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,7168,0.020618667205174766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,5120,0.007914666831493378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,5120,0.02285760045051575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,6144,0.018683733542760213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,4096,0.00680213322242101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,4096,0.021817600727081297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,5120,0.016973867019017538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,3584,0.006274133423964183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,3584,0.020734934012095134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,4096,0.014965333541234336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,3072,0.005579733351866404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,3072,0.02065066695213318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,3584,0.013713066776593527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,2560,0.005019733309745788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,2560,0.01944640080134074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,768,65536,0.13027199904123943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,2048,0.004853333532810211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,3072,0.012495999534924824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,2048,0.01871466636657715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,1536,0.004233600199222564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,2560,0.011637333035469054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,1536,0.017915733655293784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,1024,0.0035797332723935447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,1024,0.017014400164286295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,2048,0.010863999525705974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,1536,0.009947733084360758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,768,0.00306986669699351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,768,0.016530133287111917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,512,0.002571733295917511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,512,0.014888532956441245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,1024,0.008795733253161114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,256,0.0026335999369621276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,768,0.00879253347714742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,256,0.015160533785820007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,128,0.0022986667851607008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,512,0.007530666887760162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,256,0.007570133109887441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,64,0.0022709332406520844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,128,0.014664533734321594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,128,0.007035733262697856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,512,32,0.002669866631428401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,64,0.014646400014559427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,65536,0.02092906634012858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,512,32,0.01444906691710154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,16384,0.010804266730944315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,65536,0.07182719707489013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,16384,0.030806400378545123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,12288,0.008771199981371562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,12288,0.02728213270505269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,12288,0.029612799485524494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,10240,0.008359466989835102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,10240,0.026953599850336712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,16384,0.037588266531626384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,8192,0.007627733548482259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,8192,0.026444800694783527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,10240,0.025939200321833295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,7168,0.0071936001380284624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,7168,0.02472213308016459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,8192,0.022750933965047203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,6144,0.0067775999506314594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,6144,0.023545600970586143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,7168,0.020242132743199668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,5120,0.007006933291753133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,5120,0.02173866629600525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,6144,0.01908373236656189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,4096,0.00639466643333435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,4096,0.021062399943669638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,5120,0.016899200280507405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,3584,0.005865600208441416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,3584,0.020948266983032225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,4096,0.014964266618092855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,512,65536,0.12961066563924153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,3072,0.0056415999929110205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,3072,0.01916159987449646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,3584,0.014089600245157877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,2560,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,2560,0.018633600076039633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,3072,0.012852266430854797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,2048,0.004300799965858459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,2048,0.018652800718943277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,2560,0.01157973309357961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,1536,0.003822933385769526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,1536,0.016607999801635742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,2048,0.010410666465759277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,1024,0.0034730667869249977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,1536,0.00956160028775533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,1024,0.015995732943216958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,768,0.002985599885384242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,768,0.016055466731389363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,1024,0.008753066261609394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,512,0.002584533393383026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,768,0.008308266599973042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,512,0.016657066345214844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,256,0.0024746666351954144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,256,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,512,0.007895466685295106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,128,0.0021759999295075734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,256,0.007909333209196727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,128,0.006704000135262807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,128,0.01420906682809194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,64,0.0022133332987626392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,256,32,0.0023391999304294584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,64,0.01409280002117157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,256,32,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,65536,0.019155200322469076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,65536,0.06625386476516723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,16384,0.008078933258851369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,16384,0.028919466336568195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,12288,0.007160533467928569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,12288,0.02685760060946147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,16384,0.037006934483846024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,256,65536,0.12773226896921794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,12288,0.029657600323359172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,10240,0.006698666512966156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,8192,0.005919999877611796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,10240,0.02722240090370178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,8192,0.0260863999525706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,7168,0.006668800115585327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,7168,0.02484053373336792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,10240,0.026062933603922527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,6144,0.00631573349237442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,8192,0.022286933660507203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,6144,0.02313386599222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,7168,0.02059946656227112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,5120,0.0064064001043637585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,5120,0.021938133239746093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,4096,0.0059349333246548975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,4096,0.02065599958101908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,6144,0.018997333447138467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,3584,0.005713066458702088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,5120,0.01699519952138265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,3584,0.020781866709391274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,3072,0.005541333556175232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,4096,0.01493013302485148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,3072,0.01933120091756185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,2560,0.00508480022350947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,3584,0.013671466708183288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,2560,0.018823466698328652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,2048,0.004412800073623657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,3072,0.012408533692359924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,2048,0.01916159987449646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,2560,0.011621333161989848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,1536,0.003931733220815659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,1536,0.017171200116475424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,2048,0.010444800059000652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,1024,0.003047466774781545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,1024,0.015989333391189575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,1536,0.00965013305346171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,768,0.0029813334345817565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,1024,0.0083146666487058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,768,0.016531200210253397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,512,0.0025813333690166474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,512,0.016268799702326454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,768,0.008057599763075511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,256,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,512,0.007952000200748443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,256,0.014674133062362671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,128,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,256,0.007502933343251546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,128,0.01425386667251587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,64,0.0021269333859284716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,128,0.007117866476376851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,128,32,0.002209066599607468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,64,0.013386666774749756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,65536,0.01288746694723765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,128,32,0.013199999928474426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,16384,0.006275199850400289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,12288,0.005830400188763936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,16384,0.02883946696917216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,65536,0.06015359958012899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,384,128,65536,0.12674986521402995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,10240,0.005892266829808554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,12288,0.026821333169937133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,8192,0.0058037335673968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,10240,0.02553173303604126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,7168,0.0054517333706219995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,6144,0.005883733431498209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,8192,0.024939733743667602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,7168,0.02472320000330607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,5120,0.0058773333827654515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,6144,0.022780799865722658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,4096,0.0054666668176651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,3584,0.005167999863624572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,5120,0.022629332542419434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,3072,0.005041066805521647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,4096,0.021217066049575805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,2560,0.005093333125114441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,3584,0.021496532360712688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,2048,0.005038933455944061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,3072,0.01891626715660095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,1536,0.0037920000652472176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,2560,0.01861013372739156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,1024,0.0030218665798505146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,2048,0.017911465962727864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,768,0.0030389333764712016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,1536,0.01649066706498464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,512,0.002625066787004471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,1024,0.016411733627319337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,768,0.016375466187795003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,256,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,128,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,512,0.01527253290017446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,256,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,64,0.0021717332303524016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,128,0.014104533195495605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,64,32,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,65536,0.021844265858332317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,64,0.014430933197339377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,64,32,0.013579733173052468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,16384,0.007062399884064992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,12288,0.006289066871007283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,65536,0.06005866527557373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,16384,0.02879253427187602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,10240,0.0058783998092015585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,12288,0.026977066198984784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,8192,0.005884799857934316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,10240,0.025242666403452556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,7168,0.005487999816735586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,6144,0.005899733304977417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,8192,0.02478826642036438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,5120,0.00591786652803421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,7168,0.024540799856185912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,4096,0.005482666691144307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,6144,0.022759467363357544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,3584,0.00508480022350947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,5120,0.023005867004394533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,3072,0.005115733544031779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,4096,0.021912533044815063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,2560,0.005097599824269613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,3584,0.020140800873438516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,3072,0.01912320057551066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,2048,0.004665599763393402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,1536,0.0038133333126703895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,2560,0.01871573328971863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,1024,0.003047466774781545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,2048,0.018618667125701906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,1536,0.01722453236579895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,768,0.0029728000362714132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,512,0.002631466587384542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,1024,0.01602133313814799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,256,0.0025205334027608235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,768,0.015590399503707886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,128,0.002149333308140437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,512,0.014717866977055868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,64,0.0021920000513394673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,256,0.0160480002562205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,384,32,32,0.0022079999248186748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,128,0.014418133099873862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,64,0.014605866869290671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,384,32,32,0.01383039951324463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,16384,0.2851946512858073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,16384,0.5116607983907063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,12288,0.3763178825378418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,12288,0.22369279861450195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,10240,0.2744629224141439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,10240,0.15650879542032878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,8192,0.2083797295888265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,8192,0.12674986521402995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,8192,0.08093760013580323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,7168,0.17900266647338867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,7168,0.1130560000737508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,10240,0.09857280254364013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,6144,0.15757546424865723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,6144,0.09975039958953857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,12288,0.1164031982421875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,5120,0.13162559668223064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,5120,0.08764266967773438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,6144,0.06239039897918701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,4096,0.10337599913279216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,4096,0.07441919644673665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,7168,0.07117119630177816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,3584,0.09151253700256348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,3584,0.06551466782887777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,5120,0.05425386826197306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,3072,0.07966187000274658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,3072,0.05878826777140299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,16384,0.14904747009277344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,2560,0.06742719809214273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,2560,0.05267413457234701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,3584,0.04177279869715373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,2048,0.05594240029652914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,2048,0.045797332127888994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,4096,0.046453332901000975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,1536,0.042650667826334636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,1536,0.03718506495157878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,2560,0.03299946586290996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,1024,0.03097066680590312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,1024,0.03131519953409831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,3072,0.036950401465098065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,768,0.022664533058802287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,768,0.027040000756581622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,1536,0.023833600680033366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,512,0.016754132509231568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,512,0.024740266799926757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,2048,0.028886399666468304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,256,0.012492799758911132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,256,0.023269333442052207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,768,0.01709866722424825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,128,0.010063999891281128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,128,0.022166399161020915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,256,0.012993066509564718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,64,0.009574400385220845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,64,0.022992000977198283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,65536,32,0.01081706682840983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,128,0.011681066950162251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,65536,32,0.02690453330675761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,65536,0.3950655937194824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,65536,0.2531029383341471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,16384,0.10680960019429524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,16384,0.07936960061391195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,1024,0.01884053349494934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,12288,0.07947519620259604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,12288,0.06739520231882731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,65536,512,0.014747732877731323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,10240,0.06669759750366211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,10240,0.058423467477162684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,10240,0.031174399455388385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,8192,0.056987734635670986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,8192,0.04982293446858724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,12288,0.034991999467213944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,7168,0.0477781335512797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,7168,0.04588693380355835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,16384,0.04436373313268026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,6144,0.043449600537618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,6144,0.0420906662940979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,8192,0.025779199600219727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,5120,0.03600320021311442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,5120,0.037034666538238524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,7168,0.023572266101837158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,4096,0.030282666285832722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,4096,0.033029333750406904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,6144,0.020947200059890748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,3584,0.02635626594225566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,3584,0.030179200569788615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,5120,0.018273067474365235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,3072,0.023308799664179484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,3072,0.02917120059331258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,4096,0.015871999661127727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,2560,0.02035520076751709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,2560,0.025909332434336345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,2560,0.012556800246238708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,2048,0.016537599762280784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,2048,0.022977066040039063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,3072,0.013639466961224875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,1536,0.012883200248082479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,1536,0.02207466761271159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,3584,0.01529813309510549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,1024,0.010135466853777569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,1024,0.01998400092124939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,1024,0.008849066495895386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,768,0.008593066533406576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,768,0.019223467508951823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,1536,0.01002346674601237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,512,0.006445866823196411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,512,0.01716053287188212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,768,0.008814932902654012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,256,0.004678399860858917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,2048,0.011449600259462994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,256,0.016646400094032288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,128,0.004321066538492838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,512,0.00790826678276062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,128,0.017190400759379068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,64,0.003984000037113826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,64,0.016774400075276693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,16384,32,0.005177600185076395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,16384,32,0.01746986707051595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,65536,0.26539519627888997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,128,0.007276799778143566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,65536,0.1940106709798177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,16384,0.07599039872487387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,256,0.007960533102353413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,16384,0.06251946687698365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,12288,0.05635840098063151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,12288,0.054871467749277744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,12288,0.03277013301849365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,10240,0.047406931718190506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,10240,0.047996799151102706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,16384,0.04098026752471924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,8192,0.039655466874440506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,8192,0.04162240028381348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,10240,0.02895893255869548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,7168,0.034167468547821045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,7168,0.038354134559631346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,8192,0.02359573245048523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,6144,0.03061760067939758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,6144,0.03456639846165975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,7168,0.021681066354115805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,5120,0.02612053354581197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,5120,0.031249066193898518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,6144,0.019479467471440633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,4096,0.02144213318824768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,4096,0.027646932999293012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,5120,0.017854932943979898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,3584,0.01899306575457255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,3584,0.026848000288009644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,4096,0.015421866377194723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,3072,0.0165173331896464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,3072,0.024951465924580894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,3584,0.01454080045223236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,2560,0.014325333635012307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,2560,0.024200532833735147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,3072,0.013243732849756875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,2048,0.012148267030715943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,2048,0.022106667359670006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,2560,0.012486400206883748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,1536,0.010065066814422607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,1536,0.02067520022392273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,1536,0.010149332880973815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,1024,0.008045866588751475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,1024,0.0202890674273173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,2048,0.010811733206113179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,768,0.006618666648864746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,768,0.016548267006874083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,768,0.008344533046086629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,512,0.005054933329423269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,512,0.016774400075276693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,1024,0.008743466933568318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,256,0.004265599946180979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,256,0.01648319959640503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,256,0.007894399762153625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,128,0.004159999887148539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,128,0.016445866227149962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,128,0.0074986666440963745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,64,0.0038421332836151125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,64,0.015897599856058757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,12288,32,0.004241066674391428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,12288,32,0.0164682666460673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,65536,0.23056000073750815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,512,0.007915733257929485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,65536,0.20422186851501464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,16384,0.06436479886372884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,16384,0.05698239803314209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,16384,0.03943253358205159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,12288,0.04821226596832275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,12288,0.05008320013682047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,12288,0.03162986636161804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,10240,0.04108373324076335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,10240,0.04498453140258789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,16384,65536,0.15693440437316894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,8192,0.034541865189870194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,8192,0.03843306700388591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,10240,0.02765760024388631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,7168,0.029581866661707562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,7168,0.03510826826095581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,8192,0.023204267024993896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,6144,0.027137066920598345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,6144,0.031231999397277832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,12288,65536,0.13888959884643554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,5120,0.022593067089716593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,5120,0.028884265820185346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,7168,0.02114560008049011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,4096,0.01861226757367452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,4096,0.026958932479222614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,6144,0.01915093262990316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,3584,0.016530133287111917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,3584,0.024855466683705647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,5120,0.017453867197036742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,3072,0.014103466272354126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,3072,0.02353066603342692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,4096,0.015475199619928996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,2560,0.012654933333396911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,2560,0.022729599475860597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,3584,0.014569600423177084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,2048,0.011159466703732808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,2048,0.02078826626141866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,2560,0.012139733632405598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,1536,0.00907306671142578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,1536,0.018821332852045694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,3072,0.012863999605178833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,1024,0.006996266543865204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,1024,0.016827734311421712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,2048,0.01070186694463094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,768,0.005869866907596588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,768,0.016924800475438435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,1536,0.009751466910044353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,512,0.004717866579691568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,512,0.017790933450063072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,1024,0.008797867099444072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,256,0.003910399973392487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,256,0.015347199638684592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,256,0.007484800120194752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,128,0.0034656000634034475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,128,0.016038399934768677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,128,0.007147733370463054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,64,0.0034048000971476236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,512,0.007999999821186066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,64,0.015468800067901611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,10240,32,0.003509333233038584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,10240,32,0.016577066977818807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,768,0.008381866415341695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,65536,0.2011573314666748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,16384,0.05723520119984945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,65536,0.17987093925476075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,16384,0.05216639836629232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,12288,0.0420085350672404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,12288,0.04612480004628499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,12288,0.030778666337331135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,10240,0.03640640179316203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,10240,0.04080319801966349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,16384,0.03863146702448527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,8192,0.030430932839711506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,8192,0.035046398639678955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,10240,0.026477867364883424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,7168,0.02616853316624959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,7168,0.03127359946568807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,8192,0.022964266935984294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,6144,0.02375040054321289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,6144,0.030009599526723225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,7168,0.020959999163945517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,5120,0.01951040029525757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,5120,0.027229867378870648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,6144,0.019170133272806804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,4096,0.016387200355529784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,4096,0.025536000728607178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,10240,65536,0.1343018690745036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,3584,0.014412800470987955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,3584,0.022951465845108033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,5120,0.017309866348902383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,3072,0.012807466586430869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,3072,0.022899200518925987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,3584,0.013738666971524557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,2560,0.01145919958750407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,2560,0.020776534080505372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,4096,0.015372799833615622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,2048,0.009921066959698995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,2048,0.02030186653137207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,3072,0.012476799885431926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,1536,0.0074879998962084455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,1536,0.01753066579500834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,2560,0.011335466305414836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,1024,0.005538133283456167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,1024,0.016612266500790916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,2048,0.010893866419792175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,768,0.004747733473777771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,768,0.01809920072555542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,1536,0.009607467055320739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,512,0.00382080003619194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,512,0.01657386620839437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,1024,0.008843732873598735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,256,0.00346666673819224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,256,0.014653866489728292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,512,0.00792746643225352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,256,0.0075328002373377485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,768,0.0085098663965861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,128,0.0030410667260487873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,64,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,128,0.01595946649710337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,64,0.014754133423169455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,8192,32,0.0034143999218940735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,8192,32,0.01632213294506073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,65536,0.17948586146036785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,65536,0.1609557310740153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,128,0.007143466671307881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,16384,0.04988053242365519
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,16384,0.049081599712371825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,12288,0.03813759883244832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,12288,0.043852798144022626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,12288,0.030705066521962483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,10240,0.03280319968859355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,10240,0.0382368008295695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,16384,0.038341331481933597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,8192,0.02683519919713338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,8192,0.03261013428370158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,10240,0.02645653287569682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,7168,0.023414399226506552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,7168,0.0303274671236674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,8192,0.022765866915384927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,6144,0.020862932999928793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,6144,0.02844160000483195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,7168,0.02119040091832479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,5120,0.018247467279434205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,5120,0.027798400322596235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,6144,0.019041067361831664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,4096,0.014354133605957031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,4096,0.024451200167338052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,5120,0.017413334051767985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,3584,0.013123200337092081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,3584,0.02294506629308065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,4096,0.01488746702671051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,3072,0.01146986683209737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,3072,0.021769599119822184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,3584,0.014062933127085366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,2560,0.010409599542617798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,2560,0.020555732647577922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,3072,0.01241386632124583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,2048,0.008835200468699138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,2048,0.018617600202560425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,2560,0.011725866794586181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,1536,0.007131733496983846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,1536,0.0189301331837972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,2048,0.01037013332049052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,1024,0.004659200211366018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,1024,0.016796799500783284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,1024,0.008795733253161114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,768,0.00420906643072764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,768,0.01693120002746582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,1536,0.01002346674601237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,512,0.003921066721280416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,512,0.017268266280492148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,512,0.007985066870848339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,256,0.003509333233038584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,256,0.014948266744613647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,768,0.007938133180141449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,128,0.0030016000072161358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,128,0.014501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,128,0.007063466807206471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,64,0.0026975999275843303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,64,0.014593066771825156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,256,0.0076341331005096436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,7168,32,0.00344106654326121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,7168,32,0.015505066514015198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,65536,0.1642240047454834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,65536,0.1362496058146159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,16384,0.045602134863535565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,16384,0.04747200012207031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,8192,65536,0.13016106287638346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,12288,0.03534080187479655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,12288,0.04158080021540324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,16384,0.03750826517740886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,10240,0.030008532603581745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,10240,0.03505920171737671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,12288,0.030253867308298748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,8192,0.02507946689923604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,8192,0.031048534313837688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,10240,0.026419200499852497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,7168,0.021272534132003786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,7168,0.029095466931660968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,8192,0.02288853327433268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,6144,0.019246933857599895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,6144,0.02723306616147359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,6144,0.01920106609662374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,5120,0.015979733069737753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,5120,0.025303467114766436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,7168,0.02068586746851603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,4096,0.013316266735394797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,4096,0.02345493237177531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,4096,0.014490666985511779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,3584,0.011901866396268208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,3584,0.022937599817911783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,5120,0.017035732666651406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,3072,0.010877866546312969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,3072,0.02099626660346985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,3072,0.012970667084058127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,2560,0.009559466441472372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,3584,0.013277866442998252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,2560,0.018846933046976724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,2048,0.00844266712665558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,2048,0.018951465686162315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,2048,0.01086186667283376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,1536,0.006681600213050842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,1536,0.017465599377950034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,2560,0.011206400394439698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,1024,0.004814933240413666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,1024,0.01797119975090027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,1024,0.008779733379681905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,768,0.004604800045490265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,768,0.01657600005467733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,1536,0.009761066238085429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,512,0.004273066421349844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,512,0.01662506659825643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,7168,65536,0.12855040232340495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,256,0.0035616000493367515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,256,0.01470186710357666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,768,0.008338133494059246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,128,0.003339733431736628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,128,0.014539733529090881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,256,0.007542400062084198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,64,0.0030879999200503034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,128,0.007106133302052816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,6144,32,0.0033781332274278007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,64,0.014614400267601014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,512,0.007944533228874206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,6144,32,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,65536,0.13741440773010255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,65536,0.13715093930562336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,16384,0.03784000078837077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,16384,0.043295999368031815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,12288,0.029231999317804975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,12288,0.036532266934712725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,12288,0.029702399174372358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,10240,0.024768000841140746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,10240,0.03391146659851074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,16384,0.03723306655883789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,8192,0.02048106590906779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,8192,0.029053866863250732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,10240,0.02643946607907613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,7168,0.017777067422866822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,7168,0.027536000808080035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,8192,0.02249280015627543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,6144,0.01590079963207245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,6144,0.02648959954579671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,7168,0.020525866746902467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,5120,0.013350400328636169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,5120,0.02463360031445821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,6144,0.018310399850209554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,4096,0.011457066734631855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,4096,0.02286720077196757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,5120,0.016237866878509522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,3584,0.010258133212725323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,3584,0.022733867168426514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,4096,0.014144000411033631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,3072,0.009305600325266521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,3072,0.020826667547225952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,3584,0.013294933239618936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,2560,0.00827946662902832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,2560,0.018595200777053834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,3072,0.012647466858228049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,2048,0.007458133498827617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,2048,0.019045333067576088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,2560,0.01157866617043813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,1536,0.005929600199063619
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,1536,0.017755732933680216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,1536,0.00956053336461385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,1024,0.0047082667549451195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,1024,0.01726186672846476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,2048,0.010558933019638062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,768,0.004333866635958353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,768,0.01606186628341675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,1024,0.008733866612116496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,512,0.0038015998899936674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,512,0.014680533607800802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,768,0.008201600114504496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,256,0.0034229333202044168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,256,0.015901866555213928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,512,0.007898666461308797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,128,0.003011200080315272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,128,0.015256533026695251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,256,0.007613866527875264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,64,0.002733866622050603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,64,0.014614400267601014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,5120,32,0.0030613332986831666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,5120,32,0.014891733725865683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,6144,65536,0.12760106722513836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,128,0.007074133555094402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,65536,0.10911786556243896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,65536,0.10722346305847168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,16384,0.03741546471913655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,16384,0.041611735026041666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,12288,0.029022934039433797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,12288,0.0358240008354187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,12288,0.029820799827575684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,10240,0.024683733781178795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,10240,0.032689066727956136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,16384,0.037079465389251706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,8192,0.020388267437616982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,8192,0.02731093366940816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,10240,0.026405332485834758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,7168,0.01771519978841146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,7168,0.02676266630490621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,8192,0.022291199366251627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,6144,0.015523200233777365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,6144,0.024948267141977946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,7168,0.02071146567662557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,5120,0.0134442667166392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,5120,0.023577600717544556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,6144,0.018242132663726807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,4096,0.009970133503278095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,4096,0.022446932395299275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,5120,0.016685867309570314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,3584,0.00918293297290802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,3584,0.021502933899561563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,4096,0.01406826674938202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,3072,0.008172800143559773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,3072,0.019845332702000937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,3584,0.013498666882514953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,2560,0.007464533547560374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,2560,0.019014400243759156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,3072,0.012428800264994305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,2048,0.006286933521429698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,2048,0.019367466370264687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,2560,0.011704533298810323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,1536,0.005208533505598704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,1536,0.016752000649770102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,1536,0.009665066997210186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,2048,0.010444800059000652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,1024,0.0042026668787002565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,1024,0.017051732540130614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,768,0.00383146678407987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,768,0.01686826745669047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,768,0.008425600330034892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,512,0.0034602666894594826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,512,0.015513599912325541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,1024,0.008742400010426839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,256,0.0030400000512599947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,256,0.01709973414738973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,256,0.0074890668193499255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,512,0.007886933286984761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,128,0.003035733352104823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,128,0.015927466750144958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,64,0.002648533384005229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,64,0.015779200196266174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,4096,32,0.0030623999734719592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,4096,32,0.014885333180427552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,128,0.007121066749095917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,65536,0.10116266409556071
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,65536,0.10470080375671387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,16384,0.02869439919789632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,16384,0.037010133266448975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,5120,65536,0.12755946318308514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,12288,0.02221226692199707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,12288,0.03254400094350179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,16384,0.03696320056915283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,10240,0.018888533115386963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,10240,0.02897599935531616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,12288,0.02981226642926534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,8192,0.01564800043900808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,8192,0.02691626747449239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,8192,0.022291199366251627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,7168,0.013808000087738036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,7168,0.02567360003789266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,10240,0.02600319981575012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,6144,0.012187733252843221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,6144,0.024757333596547446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,6144,0.018217599391937254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,5120,0.01069760024547577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,5120,0.023009065786997476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,7168,0.020227199792861937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,4096,0.00913706620534261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,4096,0.02098133365313212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,4096,0.014550399780273438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,3584,0.008388266960779826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,3584,0.020397865772247316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,5120,0.01616106629371643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,3072,0.007513600091139476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,3072,0.020121600230534872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,3584,0.013287466764450074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,2560,0.006658133367697399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,2560,0.018935465812683107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,3072,0.012476799885431926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,2048,0.00559146652619044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,2048,0.0180351992448171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,4096,65536,0.12735146681467693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,1536,0.004646400113900503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,1536,0.017800533771514894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,2048,0.01048959990342458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,1024,0.003915733347336451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,1024,0.016609066724777223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,2560,0.011596799890200297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,768,0.0034773332377274835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,768,0.01662399967511495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,1536,0.009569066762924194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,512,0.0030165334542592366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,512,0.01611199975013733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,512,0.007906133433183034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,256,0.003010133405526479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,256,0.01558080017566681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,768,0.007986133297284443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,128,0.0026122666895389557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,128,0.015955199797948204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,1024,0.008355200290679932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,64,0.0026975999275843303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,128,0.007525333265463512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,256,0.007467733323574066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3584,32,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,64,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3584,32,0.014672000209490457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,65536,0.09339839617411295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,65536,0.09890027046203613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,16384,0.030096000432968138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,16384,0.03876373370488485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,12288,0.021205333868662517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,12288,0.031152000029881794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,12288,0.029495465755462646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,10240,0.01788160006205241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,10240,0.029127466678619384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,16384,0.03714453379313151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,8192,0.014892799655596414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,8192,0.026868265867233277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,10240,0.02597759962081909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,7168,0.012905599673589072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,7168,0.02523946762084961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,8192,0.022333866357803343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,6144,0.01169706682364146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,6144,0.024361600478490196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,7168,0.020243199666341145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,5120,0.010086400310198466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,5120,0.021988266706466676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,6144,0.018304000298182167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,4096,0.008933333555857341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,4096,0.021155200401941934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,5120,0.016133333245913185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,3584,0.007923200229803721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,3584,0.02044373353322347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,4096,0.014572800199190775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,3072,0.007285333176453908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,3072,0.02022613286972046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,3072,0.012498133381207784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,3584,0.01328000028928121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,2560,0.006333866715431213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,2560,0.01914986570676168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,2048,0.005589333176612854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,2048,0.018549333016077675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3584,65536,0.12718506654103595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,1536,0.004715733230113983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,1536,0.017155200242996216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,2048,0.010391466816266378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,1024,0.0038304001092910765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,1024,0.017267199357350667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,2560,0.01125333309173584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,768,0.003472000112136205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,768,0.016371200482050575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,768,0.008298666775226593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,512,0.0030421334008375804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,1536,0.009592533111572266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,512,0.015904000401496886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,256,0.00288426677385966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,1024,0.008323200047016144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,256,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,128,0.002628266563018163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,128,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,128,0.007118933399518331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,256,0.0075573335091273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,64,0.002586666742960612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,3072,32,0.0026538667579491934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,64,0.014452266693115234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,3072,32,0.014663466811180114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,512,0.007940266529719036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,65536,0.07328960100809732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,16384,0.025316266218821208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,65536,0.08869120279947916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,16384,0.0362175981203715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,12288,0.021014400323232017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,12288,0.030903466542561847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,12288,0.029224532842636108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,10240,0.018308266003926595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,10240,0.028832000494003297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,16384,0.03689599831899007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,8192,0.014087466398874917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,8192,0.025893332560857137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,10240,0.025561600923538208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,7168,0.01228053371111552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,7168,0.02464853326479594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,8192,0.0222762664159139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,6144,0.011390933394432068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,6144,0.02291626731554667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,7168,0.019809067249298096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,5120,0.009644800424575805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,5120,0.022111999988555908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,6144,0.018225065867106118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,4096,0.008413867155710856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,4096,0.020623999834060668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,5120,0.01658560037612915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,3584,0.0076447998483975725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,3584,0.019988266626993816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,4096,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,3072,0.006985599795977275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,3072,0.01908906698226929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,3584,0.013673599561055502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,2560,0.005907199780146281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,2560,0.018549333016077675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,3072,0.01244586706161499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,2048,0.005420800050099691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,2048,0.0172650674978892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,2560,0.0116266667842865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,1536,0.0046304002404212955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,1536,0.018078933159510292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,2048,0.010392533739407857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,1024,0.003874133278926214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,1024,0.01660053332646688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,1536,0.009553066889444987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,768,0.003454933315515518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,768,0.015140266219774882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,768,0.008276266853014629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,512,0.0030623999734719592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,1024,0.008732799688975017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,512,0.015960533420244852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,256,0.0026176000634829206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,256,0.014678399761517844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,256,0.007528533538182576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,128,0.0025653332471847535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,512,0.007934933404127757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,128,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,128,0.007147733370463054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,64,0.0025770666698614756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,64,0.014465066790580749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2560,32,0.002587733417749405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2560,32,0.01446399986743927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,65536,0.06213866472244263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,65536,0.08198613325754801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,3072,65536,0.1272320032119751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,16384,0.020891733964284263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,16384,0.035018666585286455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,12288,0.018665599822998046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,12288,0.029268266757329305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,12288,0.0288810670375824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,10240,0.015084800124168397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,10240,0.02744106650352478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,16384,0.03681173324584961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,8192,0.012974933783213297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,8192,0.025254400571187337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,10240,0.02515520056088765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,7168,0.011454932888348897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,7168,0.024387200673421226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,8192,0.021643733978271483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,6144,0.010385066270828247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,6144,0.022999467452367146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,7168,0.019849600394566853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,5120,0.009164800246556599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,5120,0.0212991992632548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,6144,0.017898666858673095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,4096,0.007879466811815897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,4096,0.02185493310292562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,5120,0.016151466965675355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,3584,0.007366399963696797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,3584,0.01917333404223124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,3584,0.013240533073743186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,3072,0.00625600020090739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,3072,0.018663465976715088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,4096,0.014318933089574179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,2560,0.005234133203824361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,2560,0.019091200828552247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,2560,0.011672533551851908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,2048,0.004733866453170777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,2048,0.016790399948755898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,3072,0.01206826666990916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,1536,0.0042581334710121155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,1536,0.01663253307342529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,1536,0.009527466694513957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,1024,0.0034272000193595886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,1024,0.015225600202878317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,2048,0.010499200224876404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,768,0.003107200066248576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,768,0.015432533621788026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,1024,0.008701866865158081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,512,0.0030026666820049284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,512,0.014615466197331747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,768,0.008423466483751934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,256,0.002705066651105881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,512,0.00795839975277583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,256,0.014552533626556396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,128,0.0026581334571043652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2560,65536,0.12741653124491376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,128,0.01585919956366221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,64,0.0026591998835404714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,256,0.007449600100517273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,64,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,2048,32,0.002584533393383026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,128,0.007187200089295705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,2048,32,0.014591999848683677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,65536,0.04771519899368286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,16384,0.01824000080426534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,65536,0.07606613636016846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,16384,0.033012266953786215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,12288,0.014973866939544677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,12288,0.03061973253885905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,12288,0.028836266199747725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,10240,0.013665067156155905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,10240,0.026998400688171387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,16384,0.03645439942677815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,8192,0.012719999750455221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,8192,0.025043199459711712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,10240,0.025118933121363325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,7168,0.011006933450698853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,7168,0.022773333390553794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,8192,0.021596799294153847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,6144,0.010121599833170573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,6144,0.02284160057703654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,7168,0.019875200589497884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,5120,0.008589866757392883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,5120,0.02097813288370768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,6144,0.018313600619633993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,4096,0.0071370666225751235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,4096,0.020742400487263998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,5120,0.01616853376229604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,3584,0.006322133541107178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,3584,0.01926400065422058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,3584,0.013269333044687906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,3072,0.005899733304977417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,3072,0.019502933820088705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,4096,0.014192000031471252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,3072,0.012452266613642375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,2560,0.0054848000407218935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,2560,0.018563199043273925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,2048,0.0053610667586326596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,2048,0.017638399203618368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,2560,0.011185066898663838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,1536,0.004264533519744873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,1536,0.016504533092180886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,2048,0.010525866349538168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,1024,0.0037258667250474296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,1024,0.017684266964594523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,1536,0.009532800316810608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,768,0.0032960000137488045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,768,0.01718399922053019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,1024,0.008803199728329976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,512,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,512,0.014763733744621277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,2048,65536,0.12775253454844157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,256,0.002809600035349528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,768,0.008353066444396973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,256,0.014666666587193808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,512,0.007996800045172375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,128,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,128,0.014636799693107605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,64,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,256,0.007436800003051758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1536,32,0.002570666621128718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,128,0.0071722666422526045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,64,0.015717333555221556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,65536,0.03655573527018229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1536,32,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,16384,0.013378133376439413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,65536,0.06635733445485434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,16384,0.029882667462031047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,12288,0.011374933520952861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,12288,0.028632533550262452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,12288,0.02889813383420308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,10240,0.010453333457310993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,10240,0.0265173335870107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,16384,0.036142933368682864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,8192,0.012340266505877178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,8192,0.02492693265279134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,10240,0.02550506591796875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,7168,0.010549333691596986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,7168,0.023127466440200806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,8192,0.021901865800221763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,6144,0.009691733121871948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,6144,0.02278613249460856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,7168,0.01979413429896037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,5120,0.008306133250395458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,5120,0.02070080041885376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,6144,0.018263467152913413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,4096,0.0068800002336502075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,4096,0.020863999923070274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,5120,0.016567466656366985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,3584,0.006268799801667531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,3584,0.018810667594273887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,3584,0.01328000028928121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,3072,0.00553706685702006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,3072,0.018549333016077675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,4096,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,2560,0.0051818668842315676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,2560,0.01798080007235209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,3072,0.012403200070063274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,2048,0.004632533093293508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,2560,0.011412266890207927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,2048,0.016535466909408568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,1536,0.003874133278926214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,1536,0.01686186591784159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,1536,0.009548800190289815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,1024,0.0034229333202044168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,1024,0.016165332992871602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,2048,0.010900266965230306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,768,0.0032320000231266023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,768,0.015796266992886863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,1024,0.008363733688990276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,512,0.0026186667382717133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,512,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,768,0.00843519965807597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,256,0.0026634665826956432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,512,0.007913599908351897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,256,0.014638933539390563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,128,0.0023520000278949736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,128,0.014439466595649719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,256,0.00749120016892751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,64,0.0021802666286627453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,64,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,1024,32,0.0021717332303524016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,128,0.007525333265463512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,1024,32,0.014484266440073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,65536,0.029498666524887085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,65536,0.06449813445409139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,16384,0.011878400047620138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,16384,0.027899734179178876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1536,65536,0.12721493244171142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,12288,0.010307199756304423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,12288,0.02707200050354004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,16384,0.03577173153559367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,10240,0.009220266342163086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,10240,0.026652799050013228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,12288,0.02890133261680603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,8192,0.008500267068545024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,8192,0.024295467138290405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,10240,0.025603200991948443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,7168,0.007909333209196727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,7168,0.022834134101867676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,8192,0.02145919998486837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,6144,0.007342933118343354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,6144,0.021972266832987468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,7168,0.019895466168721516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,5120,0.008037333190441132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,5120,0.02076906760533651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,6144,0.01823893388112386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,4096,0.006735999882221222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,4096,0.019578667481740315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,1024,65536,0.12693333625793457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,3584,0.006372266511122386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,3584,0.018729599316914876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,5120,0.01623253325621287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,3072,0.005542399982611338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,3072,0.01882986625035604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,4096,0.014243200421333313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,2560,0.0050335998336474095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,2560,0.017462400595347087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,3584,0.013330133756001792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,2048,0.004682666560014089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,2048,0.016758400201797485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,3072,0.012485333283742269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,1536,0.003995733211437861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,2560,0.011236266295115153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,1536,0.01667840083440145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,1024,0.00347626656293869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,1024,0.01639359990755717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,2048,0.010776533683141073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,768,0.0030122667551040648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,768,0.015152000387509666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,1536,0.00953493316968282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,512,0.0030016000072161358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,1024,0.008728532989819845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,512,0.014566399653752646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,256,0.0025781333446502686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,768,0.007973333199818928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,256,0.014331733187039694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,128,0.002233600119749705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,128,0.014833066860834757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,256,0.007816533247629803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,512,0.00790826678276062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,64,0.0022474666436513263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,768,32,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,64,0.01446399986743927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,128,0.007118933399518331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,65536,0.02400853236516317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,768,32,0.01434346636136373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,16384,0.009904000163078307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,65536,0.058413867155710855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,16384,0.02884053389231364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,12288,0.008406399687131246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,12288,0.026935466130574542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,12288,0.029305599133173627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,10240,0.010585600137710571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,10240,0.025820799668629962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,16384,0.03585919936498006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,8192,0.008749866485595703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,8192,0.024337067206700643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,10240,0.02568639914194743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,7168,0.008378666639328004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,8192,0.02188160022099813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,7168,0.023270400365193684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,6144,0.007699200014273326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,6144,0.022297600905100502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,7168,0.019444266955057778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,5120,0.007121066749095917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,5120,0.02071466644605001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,6144,0.018286933501561485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,4096,0.0064064001043637585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,4096,0.01980479955673218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,5120,0.016173866391181946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,3584,0.005881600081920624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,4096,0.014607999722162882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,3584,0.018623999754587808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,3072,0.0054293334484100345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,3072,0.018394666910171508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,3072,0.012422399719556172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,3584,0.013362133502960205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,2560,0.005046399931112925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,2560,0.01776640017827352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,768,65536,0.125383464495341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,2048,0.004689066608746847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,2048,0.017217065890630087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,1536,0.0040501333773136135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,1536,0.016660267114639284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,2560,0.011181867122650147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,1024,0.0033962666988372804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,2048,0.01079360047976176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,1024,0.017378133535385133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,768,0.0030303999781608583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,1536,0.009633066256841023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,768,0.01461013356844584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,512,0.0026410666604836782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,512,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,1024,0.008675199747085572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,768,0.008384000261624653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,256,0.0026581334571043652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,512,0.007912533481915791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,256,0.014646400014559427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,128,0.0022624000906944276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,128,0.014651733636856078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,256,0.007577600081761678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,64,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,128,0.007276799778143566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,64,0.01415786643822988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,512,32,0.0023039999107519788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,512,32,0.013507200280825296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,65536,0.015239466230074564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,65536,0.05367146730422974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,16384,0.008824533224105835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,16384,0.027501867214838667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,12288,0.007738666733105977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,12288,0.02523840069770813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,16384,0.03645439942677815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,10240,0.007259733478228251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,10240,0.026547199487686156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,12288,0.028833067417144774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,8192,0.007112533350785573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,8192,0.024654932816823325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,10240,0.025620265801747637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,7168,0.006772266825040181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,7168,0.023031467199325563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,8192,0.0214848001797994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,6144,0.006279466549555461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,6144,0.021604265769322714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,7168,0.020096000035603842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,5120,0.006781866649786632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,5120,0.021381332476933797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,6144,0.01816533406575521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,4096,0.00668693333864212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,4096,0.018885332345962524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,5120,0.01623146633307139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,3584,0.0059690664211908976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,4096,0.01444586714108785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,3584,0.01866453289985657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,3072,0.005527466535568237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,3072,0.018796799580256145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,3584,0.013251200318336487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,512,65536,0.12486506303151448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,2560,0.005005866785844167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,2560,0.016898133357365928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,2048,0.004496000210444133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,3072,0.012489599982897441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,2048,0.017944532632827758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,2560,0.011162666479746501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,1536,0.003973333289225897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,1536,0.016520532965660095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,1024,0.003403733422358831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,2048,0.010813867052396137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,1024,0.01495146652062734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,768,0.0030037333567937215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,1536,0.009659733374913533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,768,0.016100266575813295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,512,0.002584533393383026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,1024,0.008776533603668212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,512,0.016193067034085594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,256,0.002499199906984965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,768,0.008022400240103405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,512,0.00803306649128596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,256,0.01420906682809194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,128,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,256,0.007210666437943776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,128,0.014638933539390563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,64,0.002268799891074498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,128,0.00719893326361974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,256,32,0.0022976001103719073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,64,0.014524799585342408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,256,32,0.013610666990280152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,65536,0.014333867033322654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,65536,0.04633920192718506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,16384,0.00639466643333435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,16384,0.02493013342221578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,12288,0.006298666695753734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,12288,0.024804266293843587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,256,65536,0.12291200160980224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,10240,0.006443733473618825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,12288,0.028643200794855755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,16384,0.035905067125956217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,10240,0.02537386616071065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,8192,0.005891199906667074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,8192,0.02464639941851298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,7168,0.005639466643333435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,7168,0.02307413419087728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,10240,0.02515946626663208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,8192,0.021767467260360718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,6144,0.005859200159708659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,6144,0.022734934091567995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,5120,0.0063391998410224915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,7168,0.01994453271230062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,5120,0.020771199464797975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,4096,0.005938133100668589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,4096,0.02066453297932943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,6144,0.017761067549387614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,3584,0.005515733361244201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,5120,0.016241066654523215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,3584,0.018966400623321535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,4096,0.014212266604105631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,3072,0.005161599814891815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,3072,0.018231467405954997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,3584,0.013313066959381104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,2560,0.005131733417510986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,3072,0.012472533186276754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,2560,0.017489065726598106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,2048,0.005075199902057648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,2048,0.016919465859731038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,2560,0.011121066411336263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,1536,0.0038111999630928038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,2048,0.01048959990342458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,1536,0.01593706707159678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,1024,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,1024,0.015619200468063355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,1536,0.009529599547386169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,768,0.0030239999294281008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,1024,0.0083146666487058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,768,0.016254933675130208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,512,0.002632533262173335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,512,0.014556800325711569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,768,0.007573333382606506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,256,0.0022250667214393617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,256,0.014095999797185264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,512,0.007624533275763195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,128,0.002218666672706604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,65536,0.12258880138397217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,256,0.00716480016708374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,128,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,64,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,256,128,128,0.007123200098673503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,128,32,0.0021845333278179167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,64,0.014467199643452963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,65536,0.010390399893124899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,128,32,0.013978667060534158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,16384,0.006241066753864289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,12288,0.005515733361244201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,65536,0.04322346846262614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,10240,0.005847466488679251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,16384,0.02629973292350769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,8192,0.0058442667126655575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,10240,0.024794665972391765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,7168,0.0054624001185099285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,8192,0.023079466819763184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,6144,0.005487999816735586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,12288,0.025073067347208662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,7168,0.022849067052205404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,5120,0.005096533397833506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,4096,0.005538133283456167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,6144,0.022025599082310995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,3584,0.005443199972311656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,4096,0.02058666745821635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,5120,0.02150826652844747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,3072,0.005082666873931885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,3584,0.020145066579182944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,3072,0.01764799952507019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,2560,0.005272533496220907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,2048,0.005075199902057648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,2560,0.018487467368443807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,1536,0.0035605333745479585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,1024,0.0034815999368826545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,2048,0.017477333545684814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,1536,0.016293332974116007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,768,0.0030389333764712016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,512,0.002605866640806198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,256,0.0022495999932289124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,768,0.015877333283424378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,512,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,1024,0.01493119994799296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,128,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,256,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,64,0.0019402666638294857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,128,0.01474453310171763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,64,32,0.002165333429972331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,64,0.013833600282669067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,65536,0.016101333498954772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,16384,0.0067007998625437425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,12288,0.005433600147565206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,64,32,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,65536,0.04318079948425293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,16384,0.0253930668036143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,10240,0.005896533528963724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,12288,0.024733867247899374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,8192,0.005526400109132131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,10240,0.024482132991154988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,7168,0.005443199972311656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,6144,0.005427200098832448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,5120,0.005074133475621542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,7168,0.022779732942581177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,8192,0.023485867182413737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,6144,0.02216426730155945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,4096,0.004654933512210846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,3584,0.005031466484069824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,5120,0.021792000532150267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,3072,0.004693333307902018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,2560,0.004991999765237173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,4096,0.02074880003929138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,3584,0.01880853374799093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,3072,0.018683733542760213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,2048,0.004716800153255462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,2560,0.01692053278287252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,1536,0.003916800022125244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,2048,0.01662506659825643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,1024,0.003047466774781545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,1536,0.016429866353670754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,768,0.003053866575161616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,1024,0.015102932850519816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,512,0.0026848000784715016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,768,0.014846932888031007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,256,0.002325333406527837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,512,0.014697600404421488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,128,0.002243199944496155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,256,0.014465066790580749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,64,0.002218666672706604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,128,0.014402133226394654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,256,32,32,0.0022261333962281544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,64,0.014712533354759217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,256,32,32,0.013619200388590495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,16384,0.4020213445027669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,16384,0.2374944051106771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,12288,0.17894934018452963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,12288,0.3103679974873861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,10240,0.2449951966603597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,10240,0.14559466044108074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,8192,0.2002687931060791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,8192,0.12253226439158123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,8192,0.08049279848734538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,7168,0.17110080718994142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,7168,0.10896639823913574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,10240,0.09848213195800781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,6144,0.14993066787719728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,6144,0.09598399798075358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,12288,0.11763306458791097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,5120,0.1250762701034546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,5120,0.08321386973063151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,6144,0.0626581350962321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,4096,0.10150079727172852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,4096,0.07159146467844645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,7168,0.07188159624735514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,3584,0.08855040073394775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,3584,0.06254719893137614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,5120,0.0538271983464559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,3072,0.07708799839019775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,3072,0.05624853372573853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,16384,0.15778346061706544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,2560,0.06552639802296957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,2560,0.05001706679662069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,3584,0.04077226718266805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,2048,0.053592534859975186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,2048,0.043901864687601724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,4096,0.04552106857299805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,1536,0.0412341316541036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,1536,0.03823466698328654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,3072,0.03621120055516561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,1024,0.02895359992980957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,1024,0.02909226616223653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,2560,0.03207466602325439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,768,0.022881066799163817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,768,0.026124799251556398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,1024,0.01845973332722982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,512,0.016301866372426352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,512,0.022729599475860597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,1536,0.02301973303159078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,256,0.010048000017801921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,256,0.020595200856526694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,768,0.016669867436091106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,2048,0.027745066086451213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,128,0.00823466678460439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,128,0.018820265928904213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,64,0.0073183998465538025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,64,0.01888213356335958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,256,0.012232533097267151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,65536,32,0.00735040009021759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,65536,32,0.020706133047739664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,65536,0.3738464037577311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,65536,0.23235626220703126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,16384,0.10271146297454833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,16384,0.07733759880065919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,128,0.01167039970556895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,12288,0.07699519793192545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,12288,0.06315199931462606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,65536,512,0.014669866363207499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,10240,0.06493120193481446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,10240,0.053976531823476157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,10240,0.03078719973564148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,8192,0.05427840153376261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,8192,0.04694613218307495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,12288,0.03908586502075195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,7168,0.04540586471557617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,7168,0.04307946761449178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,16384,0.04362986485163371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,6144,0.03948266506195068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,6144,0.03926506837209066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,8192,0.02536960045496623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,5120,0.03389333486557007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,5120,0.03485440015792847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,7168,0.02545173366864522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,4096,0.028886399666468304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,4096,0.030902399619420366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,5120,0.018644267320632936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,3584,0.024864000082015992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,3584,0.027687466144561766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,6144,0.020933334032694498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,3072,0.021741867065429688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,3072,0.02635200023651123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,4096,0.01758506695429484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,2560,0.01898026665051778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,2560,0.02491733431816101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,2560,0.013148799538612366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,2048,0.015507200360298156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,2048,0.023001599311828613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,3072,0.013737600048383078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,1536,0.012172800302505494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,1536,0.020798933506011964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,3584,0.015046399831771851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,1024,0.009307733178138733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,1024,0.01849173307418823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,1024,0.009224533041318258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,768,0.007996800045172375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,768,0.016745599110921223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,1536,0.01025493343671163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,512,0.0058783998092015585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,512,0.016838399569193523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,768,0.008382933338483174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,256,0.004350933432579041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,256,0.01467519998550415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,512,0.007814399898052216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,128,0.0035125332574049628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,128,0.015289599696795145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,256,0.007910400132338206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,64,0.00393599991997083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,64,0.016562133034070333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,16384,32,0.0042912001411120095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,16384,32,0.015754666924476624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,2048,0.012498133381207784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,65536,0.26977173487345374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,128,0.007109333574771881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,16384,0.07535253365834554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,65536,0.20654080708821615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,16384,0.06807146867116293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,12288,0.05589439868927002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,12288,0.05516479810078939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,12288,0.032313599189122515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,10240,0.04730133215586345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,10240,0.048293332258860275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,16384,0.041016534964243574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,8192,0.040226133664449056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,8192,0.043220265706380205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,10240,0.028219733635584516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,7168,0.03346133232116699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,7168,0.036022400856018065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,8192,0.02415999968846639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,6144,0.03107733329137166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,6144,0.032986666758855185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,7168,0.021600000063578286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,5120,0.02574399908383687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,5120,0.030371199051539104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,6144,0.019666133324305217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,4096,0.021589332818984987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,4096,0.02738879919052124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,5120,0.01764799952507019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,3584,0.018687999248504637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,3584,0.02493333419164022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,4096,0.015864533185958863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,3072,0.016448000073432924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,3072,0.0243231991926829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,3584,0.014102400342623393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,2560,0.014276267091433207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,2560,0.022634667158126832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,3072,0.013517866532007853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,2048,0.012128000458081562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,2048,0.021215999126434328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,2560,0.012049067020416259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,1536,0.009578667084376017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,1536,0.018962132930755615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,2048,0.011276800433794658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,1024,0.007898666461308797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,1024,0.017167999347050985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,1536,0.009577600161234538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,768,0.006298666695753734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,768,0.016695467631022136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,768,0.007974400122960409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,512,0.004615466793378194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,512,0.01684053341547648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,1024,0.008869333068529765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,256,0.0039007999002933502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,256,0.015577600399653117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,512,0.007930666704972585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,128,0.003432533393303553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,256,0.0076682666937510175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,128,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,64,0.0030037333567937215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,64,0.014518400033315023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,12288,32,0.0032138665517171226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,12288,32,0.014871467153231302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,65536,0.2968149185180664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,128,0.007216000060240428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,65536,0.17991360028584796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,16384,0.0833077351252238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,16384,0.0643722653388977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,16384,0.03921066522598267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,12288,0.062293334801991784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,12288,0.053148798147837316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,12288,0.031037867069244385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,10240,0.05237760146458944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,10240,0.0459935983022054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,16384,65536,0.17177492777506512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,8192,0.04124053319295247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,8192,0.04030613501866658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,10240,0.027289599180221558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,7168,0.0371018648147583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,7168,0.0366048018137614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,12288,65536,0.13595627148946127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,6144,0.032253867387771605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,6144,0.03393280108769735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,8192,0.0233952005704244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,5120,0.027109332879384357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,5120,0.03157013257344564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,7168,0.020797866582870483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,4096,0.02282879948616028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,4096,0.027694932619730633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,5120,0.01770240068435669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,3584,0.019731199741363524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,3584,0.02589973409970601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,4096,0.015005866686503092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,3072,0.018104533354441323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,3072,0.023741867144902548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,6144,0.019081600507100425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,2560,0.015780267119407655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,2560,0.022731733322143555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,3584,0.014582399527231851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,2048,0.012729600071907043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,2560,0.011366400122642516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,2048,0.020850133895874024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,1536,0.009658666451772054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,1536,0.01899306575457255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,3072,0.012839466333389282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,1024,0.007178666690985362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,1024,0.01922773321469625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,1024,0.00839573343594869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,768,0.005942399799823761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,768,0.016778665781021117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,1536,0.009637332955996196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,512,0.004808533191680908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,512,0.016615466276804606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,768,0.007948799928029378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,256,0.0038719999293486277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,256,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,256,0.0074986666440963745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,128,0.00342399999499321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,128,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,512,0.008417066931724549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,64,0.0031701333820819853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,64,0.01470186710357666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,10240,32,0.0034602666894594826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,10240,32,0.015597866972287497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,128,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,65536,0.18538559277852376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,2048,0.010418132940928141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,16384,0.05301226774851481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,65536,0.15293760299682618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,16384,0.05335359970728556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,12288,0.03920213381449382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,12288,0.04378879865010579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,12288,0.03024853269259135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,10240,0.03349440097808838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,10240,0.03856106599171956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,16384,0.038251733779907225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,8192,0.028124799331029255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,8192,0.033011200030644734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,10240,0.026444800694783527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,7168,0.024142932891845704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,7168,0.029396265745162964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,8192,0.022734934091567995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,6144,0.021706666549046835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,6144,0.027185066541035967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,7168,0.02116586764653524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,5120,0.01837973395983378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,5120,0.025913600126902265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,10240,65536,0.13359893163045247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,4096,0.015195733308792115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,4096,0.02437653342882792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,6144,0.019011199474334717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,3584,0.01330560048421224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,3584,0.022760534286499025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,5120,0.01744746764500936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,3072,0.011881599823633831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,3072,0.022670932610829673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,4096,0.015054933230082192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,2560,0.010533333818117777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,2560,0.020477867126464842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,3584,0.01370133360226949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,2048,0.0091839998960495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,2048,0.018730666240056357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,2560,0.011762133240699768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,1536,0.007586133480072021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,1536,0.016825600465138753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,3072,0.012482133507728577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,2048,0.010375466942787171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,1024,0.005457066496213277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,1024,0.01765759984652201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,768,0.004939733445644379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,768,0.016194132963816325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,1536,0.009657599528630574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,512,0.003822933385769526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,512,0.01612053314844767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,1024,0.008317866424719492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,256,0.003044266750415166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,256,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,768,0.00844693382581075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,128,0.0030591999491055804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,128,0.015336533387502035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,256,0.007494399944941203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,64,0.0029994666576385496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,64,0.01446293294429779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,128,0.0071712002158164975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,8192,32,0.0030591999491055804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,8192,32,0.014616533120473226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,65536,0.17308799425760907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,512,0.0079925333460172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,16384,0.04870400031407674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,65536,0.13690667152404784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,16384,0.05673493146896362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,12288,0.0370197335879008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,12288,0.04619306723276774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,12288,0.030055467287699384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,10240,0.032153600454330446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,10240,0.03994986613591512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,16384,0.03785813252131144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,8192,0.02616746624310811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,8192,0.03535679976145427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,10240,0.026423466205596925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,7168,0.022743467489878336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,7168,0.02897706627845764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,8192,0.022346667448679605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,6144,0.02021333376566569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,6144,0.026964267094930012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,7168,0.02093120018641154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,5120,0.0171509325504303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,5120,0.024737066030502318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,6144,0.018757333358128868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,4096,0.014033066232999167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,4096,0.023423999547958374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,5120,0.017349332571029663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,3584,0.012553600470225015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,3584,0.02251840035120646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,4096,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,3072,0.011245866616566975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,3072,0.02134400010108948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,3584,0.013713066776593527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,2560,0.00983786682287852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,2560,0.019267199436823527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,3072,0.012077866991360982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,2048,0.008730666836102803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,2560,0.011655466755231221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,2048,0.017810134092966716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,1536,0.007064533233642578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,1536,0.018011732896169027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,1536,0.009581866860389709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,1024,0.005053866902987162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,1024,0.01869973341623942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,2048,0.010396800438563029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,768,0.005138133466243744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,768,0.01649386684099833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,1024,0.008387200037638346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,512,0.004211199780305227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,512,0.017138133446375527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,768,0.008344533046086629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,256,0.003385599950949351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,256,0.016126933693885803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,512,0.00790826678276062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,128,0.002959999938805898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,128,0.014791466792424521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,128,0.007063466807206471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,256,0.007493333518505096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,64,0.0026528000831604003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,7168,32,0.0029706666866938275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,64,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,7168,32,0.014507733782132468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,65536,0.1780469258626302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,65536,0.12198826471964519
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,16384,0.04771733283996582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,16384,0.04928426742553711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,8192,65536,0.12905066808064777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,12288,0.03705173333485921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,12288,0.04113173484802246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,16384,0.03771946827570598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,10240,0.030997333923975627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,10240,0.03494506676991781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,12288,0.02969493269920349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,8192,0.026088533798853557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,8192,0.030983465909957885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,10240,0.025998934110005693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,7168,0.022388267517089843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,7168,0.028913066784540815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,8192,0.022719999154408775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,6144,0.02009920080502828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,6144,0.027036799987157183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,7168,0.020598399639129638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,5120,0.01681813398996989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,5120,0.025383466482162477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,6144,0.01863893270492554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,4096,0.0137855996688207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,4096,0.02331519921620687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,5120,0.01685439944267273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,3584,0.01248426636060079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,3584,0.02103786667188009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,4096,0.014948266744613647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,3072,0.011106133460998535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,3072,0.02106879949569702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,3584,0.01304639975229899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,2560,0.009628799557685853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,2560,0.019808000326156615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,3072,0.012574932972590127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,2048,0.008365866541862488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,2048,0.018810667594273887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,2560,0.011204266548156738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,1536,0.006762666503588359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,1536,0.017416532834370932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,1536,0.009192533294359843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,1024,0.005031466484069824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,1024,0.01656000018119812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,7168,65536,0.12791999975840251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,768,0.00421973317861557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,768,0.015270400047302245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,2048,0.010532266894976298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,512,0.003918933371702829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,512,0.017794134219487508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,1024,0.008916266759236654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,256,0.0031114667654037476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,256,0.01458560029665629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,512,0.008030933141708375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,128,0.002713600049416224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,768,0.008331733445326488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,128,0.014657066265741984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,64,0.002678400029738744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,64,0.014674133062362671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,6144,32,0.002656000107526779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,6144,32,0.01479573349157969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,128,0.007109333574771881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,65536,0.11969599723815919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,65536,0.15804586410522461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,256,0.0075328002373377485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,16384,0.04757866859436035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,16384,0.052009598414103186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,12288,0.034543999036153156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,12288,0.03815040191014608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,12288,0.029674667119979858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,10240,0.029468800624211627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,10240,0.03442666530609131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,16384,0.037563733259836835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,8192,0.02435200015703837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,8192,0.030118399858474733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,10240,0.025999999046325682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,7168,0.02129279971122742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,7168,0.028251733382542926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,8192,0.022730666399002075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,6144,0.01880106727282206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,6144,0.02706666588783264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,7168,0.020441599686940513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,5120,0.015783466895421348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,5120,0.024395734071731567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,6144,0.019061332941055296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,4096,0.013238400220870972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,4096,0.02326186696688334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,5120,0.016152532895406087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,3584,0.011754666765530903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,3584,0.021687465906143188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,4096,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,3072,0.010550399621327717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,3072,0.02072746753692627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,3584,0.013290666540463767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,2560,0.009211732943852743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,2560,0.01872746745745341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,2560,0.011276800433794658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,2048,0.007996800045172375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,3072,0.012477866808573405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,2048,0.016953599452972413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,1536,0.006278400123119354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,1536,0.017783466974894205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,1536,0.009556266665458679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,1024,0.004518400132656098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,6144,65536,0.1275552034378052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,1024,0.01835306684176127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,768,0.00402453343073527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,768,0.0166101336479187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,2048,0.010461866855621338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,512,0.0038933334251244865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,512,0.015017599860827128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,768,0.008321066697438557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,256,0.003067733347415924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,256,0.0144405335187912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,1024,0.008734933535257975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,128,0.0026549334327379864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,128,0.014460800091425577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,256,0.007502933343251546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,64,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,512,0.007944533228874206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,64,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,5120,32,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,5120,32,0.014616533120473226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,128,0.007130666573842366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,65536,0.10771413644154866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,65536,0.10167040030161541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,16384,0.03447893460591634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,16384,0.04307733376820882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,12288,0.029287467400232952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,12288,0.03430933157602946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,12288,0.02977173328399658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,10240,0.024623999993006386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,10240,0.03141866723696391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,16384,0.037189332644144694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,8192,0.0201365331808726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,8192,0.02770986755688985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,10240,0.02643839915593465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,7168,0.017592533429463705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,7168,0.0269269327322642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,8192,0.02230400045712789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,6144,0.01565226713816325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,6144,0.024133332570393882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,7168,0.02029333313306173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,5120,0.013427199920018515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,5120,0.022581332921981813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,6144,0.018604799111684164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,4096,0.011262933413187664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,4096,0.02030613422393799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,5120,0.016587733229001363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,3584,0.008387200037638346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,3584,0.019407999515533448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,4096,0.014121599992116294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,3072,0.007618133227030437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,3072,0.018525866667429607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,3584,0.01339413324991862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,2560,0.006671999891599019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,2560,0.018652800718943277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,3072,0.012442666292190551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,2048,0.005790933469931285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,2048,0.01754666765530904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,2560,0.011246933539708456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,1536,0.004743466774622599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,1536,0.016787199179331462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,2048,0.010372266173362732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,1024,0.0038912000755469004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,1024,0.014642133315404256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,1024,0.008331733445326488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,768,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,768,0.016006400187810264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,1536,0.00921066701412201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,512,0.0030346666773160298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,512,0.01546346644560496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,768,0.008317866424719492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,256,0.0025973332424958544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,256,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,512,0.007970133423805236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,128,0.002980266759792964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,128,0.014551466703414917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,128,0.007133866846561432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,64,0.0025973332424958544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,256,0.007523199915885926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,4096,32,0.00262719988822937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,64,0.014605866869290671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,4096,32,0.01460693379243215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,65536,0.10115946928660076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,65536,0.09314133326212565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,16384,0.03078506588935852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,16384,0.046136534214019774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,5120,65536,0.12720533212025958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,12288,0.02900159955024719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,12288,0.03302613298098246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,16384,0.036841599146525066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,10240,0.022411733865737915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,10240,0.028995199998219805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,12288,0.029449599981307983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,8192,0.01963520050048828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,8192,0.028508800268173217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,8192,0.021908267339070638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,7168,0.01805013418197632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,7168,0.02768426736195882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,10240,0.026219733556111652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,6144,0.014074666301409402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,6144,0.02280319929122925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,6144,0.017975467443466186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,5120,0.011960533261299134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,5120,0.021364265680313112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,7168,0.019808000326156615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,4096,0.011636267105738323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,4096,0.019770665963490804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,4096,0.01411733329296112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,3584,0.010431999961535137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,3584,0.01965973377227783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,5120,0.0162581334511439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,3072,0.007249066730340321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,3072,0.01932586630185445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,3584,0.012825600306193032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,2560,0.006390400230884552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,2560,0.017520000537236534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,3072,0.012237866719563801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,2048,0.00561706672112147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,4096,65536,0.12722026507059733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,2048,0.018242132663726807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,1536,0.005850666761398315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,1536,0.017132800817489625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,2560,0.011249066392580668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,1024,0.0037834666669368743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,1024,0.01723946730295817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,1536,0.009289600451787313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,768,0.003915733347336451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,768,0.01621333360671997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,1024,0.008297599852085114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,512,0.0034122665723164878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,512,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,2048,0.010486400127410889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,256,0.003089066594839096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,256,0.015254400173823037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,256,0.007574399809042613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,128,0.002703999976317088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,512,0.007890133559703827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,128,0.015748266379038492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,768,0.008438400427500407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,64,0.0026496000587940215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3584,32,0.0026634665826956432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,64,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3584,32,0.014566399653752646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,128,0.007106133302052816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,65536,0.08015999794006348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,16384,0.029076266288757324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,65536,0.08665706316630045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,16384,0.03438400030136109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,12288,0.019803732633590698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,12288,0.029202133417129517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,12288,0.029660799105962116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,10240,0.01696853240331014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,10240,0.02722559968630473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,16384,0.036665598551432296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,8192,0.014040533701578775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,8192,0.02486293315887451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,10240,0.02598506609598796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,7168,0.012478933731714884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,7168,0.024074665705362954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,8192,0.02152000069618225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,6144,0.011268267035484314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,6144,0.02310933272043864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,7168,0.01984106699625651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,5120,0.009620267152786254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,5120,0.021463465690612794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,6144,0.017889066537221273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,4096,0.008472533027331034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,4096,0.019827200969060262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,5120,0.016210132837295534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,3584,0.007829333345095318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,3584,0.020028799772262573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,4096,0.01413653294245402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,3072,0.00688213308652242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,3072,0.018526933590571084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,3584,0.013318399588267008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,2560,0.0060245335102081295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,2560,0.017927465836207072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,3072,0.012006400028864543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,2048,0.005063466727733612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,2048,0.016857600212097167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,2560,0.0116565336783727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,1536,0.004278400043646494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,1536,0.016546133160591125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3584,65536,0.12663040161132813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,1024,0.003736533224582672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,1024,0.016547200083732606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,1024,0.008694400389989216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,1536,0.009880533814430237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,768,0.0033919999996821085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,768,0.015710933009783427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,512,0.0030122667551040648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,512,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,2048,0.010020266969998677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,256,0.0025727999707063037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,256,0.014990933736165366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,512,0.00755626658598582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,256,0.0071381335457166035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,128,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,128,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,64,0.002221866697072983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,768,0.008387200037638346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,3072,32,0.0025909334421157838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,64,0.013953066865603127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,3072,32,0.014570666352907815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,128,0.007096533477306366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,65536,0.0856544017791748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,16384,0.02592639923095703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,65536,0.0806389331817627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,16384,0.03742080132166545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,12288,0.02182933290799459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,12288,0.028849067290623982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,12288,0.029215999444325763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,10240,0.018771199385325114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,10240,0.026613332827885944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,16384,0.03710613250732422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,8192,0.013246933619181315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,8192,0.024964267015457155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,10240,0.0252074658870697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,7168,0.012012799580891926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,7168,0.024666666984558105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,8192,0.021951999266942343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,6144,0.011012267072995503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,6144,0.02331519921620687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,7168,0.019682133197784425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,5120,0.009423999985059103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,5120,0.02116480072339376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,6144,0.018224000930786133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,4096,0.008731733759244282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,4096,0.02138026754061381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,5120,0.016150400042533875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,3584,0.0077674667040507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,3584,0.019523199399312338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,4096,0.014444800217946372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,3072,0.007084799806276958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,3072,0.018720000982284546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,3584,0.01293333371480306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,2560,0.005981866518656413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,3072,0.012458667159080505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,2560,0.019805866479873657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,2048,0.00544106662273407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,2048,0.017128533124923705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,2048,0.010333866874376933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,1536,0.004642133414745331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,1536,0.017043199141820273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,2560,0.011567999919255573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,1024,0.004167466859022776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,1024,0.016782933473587038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,1024,0.008787199854850769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,768,0.0038602667550245917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,768,0.015569067001342774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,1536,0.009516800443331402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,512,0.003050666550795237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,512,0.016498133540153503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,768,0.008328533172607422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,256,0.003020799905061722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,512,0.007909333209196727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,256,0.014625066518783569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,128,0.0026410666604836782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,128,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,256,0.007503999769687653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,64,0.0025792000194390613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,128,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2560,32,0.0026133333643277483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,64,0.014460800091425577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2560,32,0.014957867066065469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,65536,0.06420160134633382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,65536,0.07322986920674643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,16384,0.021661865711212158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,16384,0.033725865681966144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,3072,65536,0.12718186378479004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,12288,0.01689599951108297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,12288,0.030751999219258624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,12288,0.029322665929794312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,10240,0.015072000026702882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,10240,0.02656853397687276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,16384,0.03677866856257121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,8192,0.013111467162768045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,8192,0.024782933791478477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,10240,0.025588266054789227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,7168,0.011310933033625285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,7168,0.022797866662343343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,8192,0.021644800901412964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,6144,0.01058240036169688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,6144,0.022913066546122234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,7168,0.01986133257548014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,5120,0.009126399954160053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,5120,0.020117332537968956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,6144,0.018423465887705485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,4096,0.007965866724650066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,4096,0.020336000124613444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,5120,0.016565333803494772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,3584,0.006974933544794719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,3584,0.018557866414388023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,4096,0.014290133118629455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,3072,0.0067893331249554946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,3072,0.01912533243497213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,3584,0.013238400220870972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,2560,0.005480533341566721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,2560,0.01789120038350423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,3072,0.01225493351618449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,2048,0.004792533318201701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,2048,0.017157334089279174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,2560,0.011183999975522359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,1536,0.004206933577855428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,1536,0.016613333423932394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2560,65536,0.12745707035064696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,1024,0.0036981334288915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,1536,0.009195733070373534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,1024,0.015490133563677469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,768,0.0035807999471823373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,768,0.015064533551534018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,2048,0.010442666212717692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,512,0.0030634666482607523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,512,0.014616533120473226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,1024,0.008709333340326945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,256,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,768,0.008022400240103405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,256,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,512,0.007642666498819987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,128,0.0026506667335828146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,256,0.007529599964618683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,128,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,64,0.002573866645495097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,64,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,2048,32,0.0026591998835404714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,128,0.007086933155854543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,2048,32,0.014759467045466105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,65536,0.0455135981241862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,65536,0.0678325335184733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,16384,0.015892266233762106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,16384,0.030953600009282427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,12288,0.012985600034395852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,12288,0.028318933645884198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,12288,0.028602667649586994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,10240,0.011916800340016683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,10240,0.025595732529958087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,16384,0.03653653462727864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,8192,0.0124917338291804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,8192,0.024409600098927817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,10240,0.025106134017308553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,7168,0.010967466235160827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,7168,0.022775467236836752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,8192,0.022388267517089843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,6144,0.009975467125574749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,6144,0.021201066176096597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,7168,0.01980479955673218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,5120,0.008333866794904072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,5120,0.02075200080871582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,6144,0.018423465887705485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,4096,0.007699200014273326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,4096,0.01981333295504252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,5120,0.016503467162450155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,3584,0.006403199831644694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,3584,0.018914133310317993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,4096,0.014132266243298849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,3072,0.0060576001803080235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,3072,0.018998400370279948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,3584,0.01323946714401245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,2560,0.00510506679614385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,2560,0.016692266861597697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,3072,0.012691199779510498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,2048,0.004671999812126159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,2560,0.011155200004577637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,2048,0.016718933979670204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,1536,0.00425600012143453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,1536,0.01665279964605967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,1536,0.00958720048268636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,1024,0.0034976000587145484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,2048,0.010506666700045268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,1024,0.016759467124938966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,768,0.00308693324526151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,768,0.014792533715566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,1024,0.008373333017031352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,512,0.0025727999707063037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,768,0.008411733309427898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,512,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,256,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,256,0.014184533556302389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,256,0.007482666770617168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,128,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,512,0.007969066500663757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,128,0.013524267077445983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,64,0.002253866692384084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,64,0.014110933740933737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1536,32,0.0022005334496498106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,128,0.007190399865309398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1536,32,0.014497066537539164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,65536,0.03657386700312297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,65536,0.05949120124181112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,16384,0.012641066312789917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,16384,0.02892906665802002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,2048,65536,0.12675626277923585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,12288,0.011160533626874287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,12288,0.027481599648793535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,12288,0.028935466210047407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,10240,0.01009493370850881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,10240,0.02486506700515747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,16384,0.03585066795349121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,8192,0.008798933029174805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,8192,0.022937599817911783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,8192,0.021418666839599608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,7168,0.008288000027338665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,7168,0.022053333123524983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,10240,0.025244800249735515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,6144,0.009773866335550944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,6144,0.020862932999928793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,7168,0.019796266158421835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,5120,0.008328533172607422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,5120,0.020278400182723998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,6144,0.017875200510025023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,4096,0.006865066786607106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,4096,0.019113600254058838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,5120,0.016179200013478598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,3584,0.006468266745408376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,3584,0.019581866264343262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,4096,0.014132266243298849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,3072,0.005570133527119955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,3072,0.019285333156585694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,3584,0.013229866822560629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,2560,0.0056085333228111265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,2560,0.01771413286526998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,3072,0.01223360002040863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,2048,0.00462719996770223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,2048,0.017465599377950034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,2560,0.01167466640472412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,1536,0.003992533435424169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,1536,0.015637333194414772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,2048,0.010427733262379963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,1024,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,1024,0.016705065965652466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1536,65536,0.12693333625793457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,768,0.0032000000278155005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,768,0.014945066968599954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,1024,0.00872320036093394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,512,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,512,0.016102400422096253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,1536,0.00959999958674113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,256,0.00262719988822937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,256,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,768,0.008478933572769165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,512,0.007793066898981731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,128,0.002430933217207591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,256,0.007574399809042613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,128,0.014115200440088908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,64,0.0022944000860055285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,1024,32,0.002295466760794322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,64,0.014446933070818582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,128,0.007225599884986877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,1024,32,0.014551466703414917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,65536,0.027628799279530842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,16384,0.014057599504788718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,65536,0.057088001569112146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,16384,0.028665600220362346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,12288,0.011819733182589214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,12288,0.026525866985321046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,12288,0.028743465741475422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,10240,0.010567466417948406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,10240,0.024692267179489136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,16384,0.03601706822713216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,8192,0.009717333316802978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,8192,0.024038400252660116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,10240,0.025175466140111284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,7168,0.00876800020535787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,7168,0.021715199947357176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,8192,0.021512534221013388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,6144,0.008078933258851369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,6144,0.02093120018641154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,7168,0.019447465737660728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,5120,0.007146666447321574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,5120,0.0206496000289917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,6144,0.018283732732137046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,4096,0.006775466601053874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,4096,0.01894186735153198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,5120,0.016102400422096253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,3584,0.005886933207511902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,3584,0.01881386637687683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,4096,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,3072,0.005566933254400889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,3584,0.013247999548912048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,3072,0.01698346734046936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,2560,0.0050346667567888895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,1024,65536,0.1256117343902588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,2560,0.01841920018196106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,2048,0.004578133424123129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,2048,0.016577066977818807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,2560,0.011262933413187664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,3072,0.012216533223787945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,1536,0.003996799886226654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,2048,0.010795733332633973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,1536,0.015441067020098367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,1024,0.003521066655715307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,1024,0.01602133313814799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,768,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,768,0.015191466609636942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,1024,0.008354133367538452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,1536,0.009556266665458679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,512,0.0026805333793163298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,768,0.008171733220418293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,256,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,512,0.01601920028527578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,256,0.013449600338935852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,128,0.00223786657055219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,128,0.013672533631324767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,512,0.007950933277606964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,256,0.007166933516661327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,64,0.0022250667214393617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,768,32,0.0022613334159056345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,128,0.006677333513895671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,64,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,768,32,0.013687466581662497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,65536,0.022460800409317017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,16384,0.011237333218256634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,65536,0.051463464895884194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,16384,0.027371732393900554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,12288,0.009567999839782714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,12288,0.024774400393168132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,12288,0.02887360056241353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,10240,0.00855466624101003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,16384,0.03617386817932129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,10240,0.0247925341129303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,8192,0.007720533510049183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,8192,0.022856533527374268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,10240,0.025312000513076784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,7168,0.00765119989713033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,7168,0.02299626668294271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,8192,0.021475199858347574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,6144,0.00668693333864212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,6144,0.02064853310585022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,7168,0.019951999187469482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,5120,0.007328000168005626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,6144,0.01776319940884908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,5120,0.02071573336919149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,4096,0.006438399851322174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,4096,0.019731199741363524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,5120,0.01575040022532145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,3584,0.00586346685886383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,3584,0.01930026610692342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,4096,0.014220800002415976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,3072,0.005612800021966299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,3072,0.017918932437896728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,3584,0.01323946714401245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,2560,0.005019733309745788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,2560,0.016872533162434897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,3072,0.012476799885431926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,2048,0.00448639988899231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,768,65536,0.12534720102945965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,2048,0.016596266627311708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,1536,0.0038272000849246977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,1536,0.0162090669075648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,2560,0.011194666226704914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,1024,0.0032501332461833954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,1024,0.015004799763361613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,1536,0.009563733140627544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,2048,0.010398933291435241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,768,0.0030303999781608583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,768,0.015783466895421348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,512,0.0026805333793163298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,1024,0.00832426647345225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,512,0.015415466825167336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,256,0.002566399921973546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,768,0.00795413355032603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,256,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,128,0.002290133386850357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,512,0.007514666517575581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,256,0.007134933272997539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,128,0.014443733294804893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,64,0.0021685334543387097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,512,32,0.0022101332743962605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,128,0.006737066805362702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,64,0.014537599682807923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,65536,0.01880000034968058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,512,32,0.014522666732470194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,65536,0.04327893257141113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,16384,0.008066133161385854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,16384,0.025803732872009277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,12288,0.007111466427644093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,12288,0.025059199333190917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,12288,0.028934399286905926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,10240,0.006693333387374878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,10240,0.02478826642036438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,16384,0.03591786623001099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,8192,0.005914666752020518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,8192,0.02286506692568461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,10240,0.025601067145665485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,7168,0.005870933334032694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,8192,0.021449599663416544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,7168,0.0229695995648702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,6144,0.006362666686375936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,6144,0.022188800573349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,7168,0.019452800353368126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,5120,0.006680533289909363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,5120,0.01903466582298279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,6144,0.017798399925231932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,4096,0.005936000247796377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,4096,0.01993173360824585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,5120,0.01623040040334066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,3584,0.005684266487757364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,4096,0.014166399836540222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,3584,0.018411733706792197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,3072,0.005470933516820272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,512,65536,0.12357119719187419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,3072,0.01739733417828878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,2560,0.005032533407211303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,2560,0.017615999778111777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,3584,0.013281066219011942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,2048,0.004282666742801667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,2048,0.016730666160583496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,3072,0.012335999806722005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,1536,0.003818666686614355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,2560,0.011218133568763732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,1536,0.017077332735061644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,1024,0.0034506666163603462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,1024,0.01460906664530436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,2048,0.010743467013041179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,768,0.002995199958483378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,1536,0.009554133812586466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,768,0.014833066860834757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,512,0.0026101333399613695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,1024,0.008757332960764568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,768,0.007899733384450276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,512,0.0149536003669103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,256,0.0022111999491850535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,256,0.014711466431617738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,512,0.00754559983809789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,128,0.002179199953873952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,256,0.007550933460394542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,128,0.013355732957522074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,64,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,128,0.007121066749095917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,64,0.014121599992116294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,256,32,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,65536,0.012905599673589072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,256,32,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,65536,0.03919893503189087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,16384,0.006355200211207073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,16384,0.026524800062179565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,256,65536,0.12268906434377033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,12288,0.00553599993387858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,12288,0.02547093431154887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,16384,0.03597653309504191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,10240,0.006051200131575266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,10240,0.02362133264541626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,8192,0.005899733304977417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,8192,0.022760534286499025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,12288,0.02844480077425639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,7168,0.005541333556175232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,10240,0.025304534037907916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,7168,0.021242666244506835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,8192,0.02152000069618225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,6144,0.005906133353710175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,6144,0.02074026664098104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,5120,0.005926399926344554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,5120,0.0207370658715566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,6144,0.017734400431315103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,7168,0.019530665874481202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,4096,0.005453866720199585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,4096,0.02066133419672648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,3584,0.005430399874846141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,5120,0.01623253325621287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,3584,0.01864746610323588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,3072,0.005172266562779745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,4096,0.014017066359519959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,3072,0.018677333990732826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,2560,0.0055071999629338585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,3584,0.013700266679128012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,2560,0.016777600844701132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,2048,0.005100800096988678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,3072,0.012025599678357441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,2048,0.017755732933680216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,1536,0.003807999938726425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,2560,0.011310933033625285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,1536,0.016583466529846193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,1024,0.0034080001215140024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,2048,0.010427733262379963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,1024,0.015105066696802774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,1536,0.009669333696365356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,768,0.002962133288383484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,1024,0.008345599969228108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,768,0.014725333452224732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,512,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,768,0.007976533472537994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,512,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,65536,0.12234773635864257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,256,0.002243199944496155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,128,0.002221866697072983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,512,0.007483733197053273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,256,0.01483519971370697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,128,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,64,0.002238933245340983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,256,0.0067114666104316715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,128,32,0.002205866575241089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,192,128,128,0.006751999755700429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,64,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,65536,0.012402133146921793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,128,32,0.014524799585342408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,16384,0.006717866659164429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,12288,0.005464533468087515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,10240,0.0058677335580190025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,65536,0.037226665019989016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,16384,0.024979199965794882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,12288,0.024699733654658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,8192,0.005858133236567179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,7168,0.0058677335580190025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,10240,0.02253440022468567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,8192,0.022526933749516805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,6144,0.005448533097902933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,5120,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,7168,0.021677867571512858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,4096,0.004645333190759023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,6144,0.020623999834060668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,5120,0.02146880030632019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,3584,0.00505920002857844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,3072,0.005086933573087057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,4096,0.02013546625773112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,2560,0.00506986677646637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,3584,0.018526933590571084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,3072,0.017912532885869345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,2048,0.005074133475621542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,1536,0.0037952000896135964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,2560,0.017350399494171144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,2048,0.017422932386398315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,1024,0.0029898665845394133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,1536,0.01606186628341675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,768,0.0029834667841593427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,1024,0.01548479994138082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,512,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,256,0.002217599997917811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,768,0.015470932920773825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,512,0.016481066743532814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,128,0.0022101332743962605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,256,0.014255999525388082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,64,0.002143999934196472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,64,32,0.0022367998957633973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,128,0.013870933651924133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,64,0.013017599781354269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,65536,0.012969600160916648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,64,32,0.014254933595657349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,16384,0.006587733328342438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,65536,0.03914346694946289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,12288,0.005885866781075796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,16384,0.02491413354873657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,10240,0.005921066800753275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,12288,0.02338026762008667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,8192,0.005916800101598104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,10240,0.022733867168426514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,7168,0.005513600011666616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,8192,0.020803199211756388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,6144,0.005138133466243744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,5120,0.005085866649945577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,7168,0.0216757337252299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,6144,0.020830933252970377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,4096,0.004713599880536398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,5120,0.02061226765314738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,3584,0.005072000126043955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,4096,0.019922133286794028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,3072,0.004746666550636292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,2560,0.0050687998533248905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,3584,0.019342933098475137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,3072,0.017800533771514894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,2048,0.0050805335243542995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,1536,0.0038176000118255613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,2048,0.017052799463272095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,2560,0.016826667388280234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,1024,0.003011200080315272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,768,0.002614400039116542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,1536,0.016404267152150473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,1024,0.014734933773676554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,512,0.0026549334327379864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,768,0.015324800213177999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,256,0.002217599997917811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,512,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,128,0.002203733225663503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,64,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,256,0.013879467050234476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,128,0.014502400159835815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,192,32,32,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,64,0.01332373321056366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,192,32,32,0.014433067043622336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,16384,0.35322453180948893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,16384,0.21020587285359701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,12288,0.261955197652181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,12288,0.16014720598856608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,10240,0.24249280293782555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,10240,0.14074986775716145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,8192,0.19335467020670574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,8192,0.11615466276804606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,8192,0.08816320101420085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,7168,0.16521600087483723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,7168,0.10276799996693928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,10240,0.10886507034301758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,6144,0.14510080019632976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,6144,0.09154559771219889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,12288,0.1202016035715739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,5120,0.12033706506093342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,5120,0.07931520144144694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,6144,0.0680405298868815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,4096,0.0976703961690267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,4096,0.06877013047536215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,7168,0.07747306823730468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,3584,0.08529919783274333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,3584,0.060651731491088864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,5120,0.058829867839813234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,3072,0.0741312026977539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,3072,0.054618668556213376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,16384,0.1765247980753581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,2560,0.06279360055923462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,2560,0.047353601455688475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,3584,0.04400533437728882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,2048,0.05247146685918173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,2048,0.04227199951807658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,4096,0.0471562663714091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,1536,0.039236267407735184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,1536,0.03574933211008708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,3072,0.03757226864496867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,1024,0.02749119997024536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,1024,0.02923626701037089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,2560,0.03320639928181966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,768,0.02176106572151184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,768,0.02507839997609456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,1536,0.023269333442052207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,512,0.015338666240374246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,512,0.022678399085998537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,2048,0.028990934292475384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,256,0.009687466422716777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,256,0.02076479991277059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,1024,0.01914773384730021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,128,0.007381333410739899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,128,0.0197653333346049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,128,0.011582932869593303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,64,0.006419200201829274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,64,0.020555732647577922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,65536,32,0.008225066463152568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,65536,32,0.022806400060653688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,65536,0.36898771921793616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,256,0.012023466825485229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,65536,0.2484842618306478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,16384,0.10283199946085612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,16384,0.07388906478881836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,768,0.01715839902559916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,12288,0.07638506889343262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,12288,0.06012586752573649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,65536,512,0.014944000045458474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,10240,0.06349759896596273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,10240,0.05205440123875936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,10240,0.033326933781305954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,8192,0.05328960021336874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,8192,0.04626133441925049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,12288,0.042328532536824545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,7168,0.04462186495463054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,7168,0.04133439858754476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,16384,0.05332266489664713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,6144,0.03899946610132853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,6144,0.0390122652053833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,8192,0.027590399980545043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,5120,0.03369706471761068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,5120,0.03425813515981038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,7168,0.028305067618687944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,4096,0.028059732913970948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,4096,0.02994026740392049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,5120,0.019869865973790486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,3584,0.02445759971936544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,3584,0.027155200640360515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,6144,0.023798400163650514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,3072,0.021732266743977866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,3072,0.025909332434336345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,4096,0.018718934059143065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,2560,0.01853440006573995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,2560,0.024474666515986124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,2560,0.014095999797185264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,2048,0.015338666240374246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,2048,0.022894932826360067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,3584,0.015590399503707886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,1536,0.012018133203188579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,1536,0.021208532651265464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,3072,0.015177599589029946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,1024,0.009246933460235595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,1024,0.019345066944758096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,1024,0.009637332955996196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,768,0.007980800171693166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,768,0.017061332861582436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,1536,0.010018133123715718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,512,0.005603200197219849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,512,0.016673066218694053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,768,0.008727467060089112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,256,0.004309333364168803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,512,0.007971199850241344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,256,0.016355199615160625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,128,0.0035135999321937563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,128,0.014711466431617738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,2048,0.01262399951616923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,64,0.00314026673634847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,64,0.016179200013478598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,16384,32,0.0034389334420363107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,128,0.007416533430417378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,16384,32,0.014643200238545737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,65536,0.2656287988026937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,65536,0.1897493362426758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,16384,0.07409280141194662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,256,0.00749120016892751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,16384,0.06232106685638428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,12288,0.05471573273340861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,12288,0.05032213528951009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,12288,0.03248320023218791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,10240,0.04689280192057292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,10240,0.044361599286397296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,16384,0.04059733152389526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,8192,0.03914133310317993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,8192,0.039206401507059736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,10240,0.028598399957021077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,7168,0.03315733273824056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,7168,0.03500906626383464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,8192,0.023172267278035483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,6144,0.02993706663449605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,6144,0.03170880079269409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,7168,0.02183253367741903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,5120,0.025399466355641682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,5120,0.028823467095692952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,6144,0.01925119956334432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,4096,0.02100906570752462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,4096,0.026013867060343428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,5120,0.01796906590461731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,3584,0.018574933211008705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,3584,0.02487679918607076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,4096,0.015312000115712484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,3072,0.01597653329372406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,3072,0.02451200087865194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,3584,0.0145578662554423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,2560,0.014166399836540222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,2560,0.02273919979731242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,3072,0.013338667154312134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,2048,0.012066133817036946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,2048,0.020983467499415077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,2560,0.01246399978796641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,1536,0.010026666522026061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,1536,0.018686934312184652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,1536,0.010188800096511842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,1024,0.007652266820271809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,1024,0.016703999042510985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,2048,0.01083733340104421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,768,0.006010666489601135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,768,0.01652906636397044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,1024,0.008849066495895386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,512,0.004740266501903534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,512,0.016114133596420287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,768,0.00836906631787618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,256,0.0035968000690142312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,256,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,512,0.008035199840863545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,128,0.0035648000737031303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,128,0.014871467153231302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,256,0.007803733150164287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,64,0.0033952000240484873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,64,0.01461120049158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,12288,32,0.003402666747570038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,12288,32,0.014713600277900696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,128,0.00764160007238388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,65536,0.2387925306955973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,65536,0.17740267117818195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,16384,0.06736319859822591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,16384,0.06437866687774658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,16384,0.03919359842936198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,12288,0.04990293184916179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,12288,0.05056106646855672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,12288,0.031035733222961426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,10240,0.04277546803156535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,10240,0.0444757342338562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,16384,65536,0.17795839309692382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,8192,0.03550613323847453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,8192,0.04197653134663899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,10240,0.027346134185791016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,7168,0.03032533327738444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,7168,0.03290560046831767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,8192,0.022834134101867676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,6144,0.02723840077718099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,6144,0.03039146661758423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,7168,0.020997333526611327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,5120,0.023749333620071412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,5120,0.027508266766866046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,12288,65536,0.14106240272521972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,4096,0.018744534254074095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,4096,0.026231465737024943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,6144,0.01907093326250712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,3584,0.01660586694876353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,3584,0.023641600211461385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,4096,0.014997333288192749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,3072,0.014947199821472168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,3072,0.023695999383926393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,5120,0.01729173262914022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,2560,0.012870400150616964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,2560,0.02268480062484741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,3584,0.013498666882514953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,2048,0.011286399761835734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,2560,0.01199679970741272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,2048,0.020551466941833497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,1536,0.009146666526794434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,1536,0.018243199586868285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,3072,0.013316266735394797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,1024,0.007009066641330719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,1024,0.018437333901723228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,1024,0.008477866649627686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,768,0.005870933334032694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,768,0.016511999567349753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,1536,0.00962453285853068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,512,0.004587733248869578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,512,0.016182399789492288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,512,0.007874133189519246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,256,0.0037770666182041167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,256,0.016676266988118492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,768,0.008409600456555684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,128,0.003370666752258936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,128,0.014985600113868713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,128,0.007167999943097432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,64,0.003049599876006444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,64,0.014802133043607077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,2048,0.010930132865905762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,256,0.0075103998184204105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,10240,32,0.003421866645415624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,10240,32,0.014578133821487427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,65536,0.17423253059387206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,65536,0.15049813588460287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,16384,0.0506933331489563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,16384,0.049379201730092366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,12288,0.03738986651102702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,12288,0.04012906551361084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,12288,0.03031146725018819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,10240,0.03214079936345418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,10240,0.03510293165842692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,16384,0.037945600350697835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,8192,0.026900267601013182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,8192,0.031193600098292036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,10240,0.026598399877548216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,7168,0.023104000091552734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,7168,0.028674133618672687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,8192,0.022529067595799764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,6144,0.020563199122746786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,6144,0.027793065706888838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,7168,0.02071466644605001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,5120,0.017534933487574258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,5120,0.025497599442799883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,10240,65536,0.13141653537750245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,4096,0.01420906682809194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,4096,0.022898133595784506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,6144,0.01875413258870443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,3584,0.012787200013796487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,3584,0.022861866156260173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,5120,0.01756160060564677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,3072,0.011316266655921937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,3072,0.02142826716105143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,3584,0.01374613344669342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,2560,0.010014933347702027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,2560,0.02127573291460673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,4096,0.014987732966740927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,2048,0.008653866251309712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,2048,0.01863893270492554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,3072,0.012882133324941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,1536,0.007237333556016285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,1536,0.017946666479110716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,2048,0.010407466689745586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,1024,0.005086933573087057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,1024,0.015939199924468996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,2560,0.0116757333278656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,768,0.0047082667549451195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,768,0.016596266627311708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,1024,0.008374399940172831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,512,0.003435733417669932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,512,0.01657386620839437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,768,0.008380800485610962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,256,0.0031199999153614043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,256,0.014630400141080222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,256,0.007586133480072021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,128,0.0027818667391935987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,128,0.015245866775512696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,1536,0.00962453285853068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,64,0.002648533384005229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,64,0.014656000336011252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,8192,32,0.0027295999228954316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,8192,32,0.014645333091417948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,512,0.007964799801508587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,65536,0.16216106414794923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,128,0.007066666583220164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,16384,0.04628693262736003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,65536,0.12758293151855468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,16384,0.05592853228251139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,12288,0.034430932998657224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,12288,0.038355199495951335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,12288,0.02985386649767558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,10240,0.029865600665410358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,10240,0.03419306675593058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,16384,0.03748586575190226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,8192,0.024810665845870973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,8192,0.029029333591461183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,10240,0.026096000274022417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,7168,0.021618133783340453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,7168,0.02823893427848816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,8192,0.022385066747665404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,6144,0.01899519960085551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,6144,0.026100265979766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,7168,0.020759467283884683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,5120,0.01623253325621287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,5120,0.024965333938598632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,6144,0.019067732493082683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,4096,0.013269333044687906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,4096,0.023034665981928507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,5120,0.017093332608540852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,3584,0.012084266543388367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,3584,0.022555732727050783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,4096,0.014577066898345948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,3072,0.010762666662534077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,3072,0.020787199338277183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,3584,0.013709867000579834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,2560,0.00953493316968282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,2560,0.01880319913228353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,3072,0.012397866447766621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,2048,0.008319999774297078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,2048,0.017089066902796428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,2560,0.011700266599655151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,1536,0.006829866766929626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,1536,0.017114667097727458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,2048,0.010613333185513813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,1024,0.004709333181381226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,1024,0.01700053413709005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,1536,0.00965013305346171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,1024,0.008712533116340637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,768,0.004391466577847799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,768,0.017832533518473307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,512,0.004243200023969015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,512,0.016642133394877114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,512,0.007492266595363617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,256,0.00340693344672521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,256,0.0145578662554423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,768,0.008157866696516674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,128,0.003049599876006444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,128,0.015108266472816467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,256,0.0075647999842961625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,64,0.002566399921973546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,64,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,7168,32,0.002980266759792964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,7168,32,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,65536,0.14714239438374838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,128,0.0075434664885203045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,65536,0.11395306587219238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,16384,0.04719786643981934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,16384,0.04756160179773967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,8192,65536,0.1284821351369222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,12288,0.03645333449045817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,12288,0.03916906515757243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,16384,0.03694506486256917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,10240,0.031035733222961426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,10240,0.03378346761067708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,12288,0.02949333389600118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,8192,0.02569813330968221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,8192,0.03039146661758423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,10240,0.026312534014383954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,7168,0.022361600399017335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,7168,0.029281065861384077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,8192,0.022457599639892578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,6144,0.020040533939997354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,6144,0.02650559941927592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,7168,0.02066133419672648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,5120,0.0166101336479187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,5120,0.02503040035565694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,6144,0.01871466636657715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,4096,0.013754666845003764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,4096,0.02245546579360962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,5120,0.016752000649770102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,3584,0.012378666798273723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,3584,0.021845332781473794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,4096,0.01411946713924408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,3072,0.011289599537849426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,3072,0.020846933126449585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,3584,0.013274666666984559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,2560,0.009595732887585957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,3072,0.012157866358757019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,2560,0.01958613395690918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,2048,0.008330666522185007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,2048,0.018713599443435668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,2048,0.010494933525721232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,1536,0.006769066552321117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,1536,0.01821546753247579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,7168,65536,0.1278197367986043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,1024,0.005109333495299021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,1024,0.017410133282343546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,2560,0.011762133240699768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,768,0.004366933306058248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,768,0.015281066298484802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,768,0.008040533463160197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,512,0.003937066594759623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,1024,0.00879253347714742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,512,0.016402133305867515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,256,0.0031114667654037476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,256,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,1536,0.00957973301410675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,128,0.0030346666773160298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,256,0.007546666761239369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,128,0.014490666985511779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,64,0.002573866645495097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,64,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,6144,32,0.002951466788848241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,6144,32,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,128,0.007153066496054332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,512,0.007561600208282471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,65536,0.16058346430460613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,16384,0.046054399013519286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,65536,0.1098965326944987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,16384,0.05221973260243734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,12288,0.03410880168279012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,12288,0.03673493464787801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,12288,0.02969493269920349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,10240,0.02943893273671468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,10240,0.03289600014686585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,16384,0.03762453397115072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,8192,0.02424959937731425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,8192,0.029121067126591998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,10240,0.025998934110005693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,7168,0.021232000986735024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,7168,0.028254934151967365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,8192,0.022552533944447836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,6144,0.01875413258870443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,6144,0.026530132691065474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,7168,0.020295466979344687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,5120,0.01565333306789398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,5120,0.025063467025756837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,6144,0.018669867515563966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,4096,0.013242666920026144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,4096,0.02254613240559896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,5120,0.016594133774439492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,3584,0.011677866180737812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,3584,0.020658133427302043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,4096,0.014625066518783569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,3072,0.010585600137710571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,3072,0.01911999980608622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,3584,0.013275733590126038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,2560,0.009239466985066731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,2560,0.018643200397491455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,2560,0.01121493379275004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,2048,0.008054399987061818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,2048,0.01808746655782064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,3072,0.012133333086967468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,1536,0.006381866832574208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,1536,0.018154666821161906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,1536,0.009690666198730468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,1024,0.004729599754015604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,1024,0.01673706571261088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,2048,0.010391466816266378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,768,0.004264533519744873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,768,0.015659733613332113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,768,0.008051200211048127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,512,0.0038442666331926978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,512,0.015686399737993875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,6144,65536,0.1272885322570801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,256,0.003253333270549774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,1024,0.008840533097585042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,256,0.01463573376337687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,128,0.003058133274316788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,128,0.014546133081118264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,512,0.007529599964618683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,256,0.0075082664688428235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,128,0.007115733126799266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,32,0.0026261332134405774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,5120,64,0.002605866640806198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,64,0.014553599556287131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,5120,32,0.01470080018043518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,65536,0.09900693098704019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,16384,0.031973334153493245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,65536,0.0948245366414388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,16384,0.03929813305536906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,12288,0.025603200991948443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,12288,0.029334400097529096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,12288,0.029633067051569623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,10240,0.019244800011316933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,10240,0.02746666669845581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,16384,0.036761601765950516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,8192,0.01635199983914693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,8192,0.027313067515691118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,10240,0.026346667607625322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,7168,0.013917866349220275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,7168,0.02498240073521932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,8192,0.021911466121673585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,6144,0.012569600343704223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,6144,0.02295893430709839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,7168,0.0200000007947286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,5120,0.0108842670917511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,5120,0.021647999684015908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,6144,0.018297600746154784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,4096,0.010714667042096455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,4096,0.02073919971783956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,5120,0.016498133540153503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,3584,0.009562666217486065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,3584,0.01903466582298279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,4096,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,3072,0.007627733548482259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,3072,0.017174400885899863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,3584,0.013221333424250284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,2560,0.006460799773534138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,2560,0.017018665870030723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,3072,0.012507733702659608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,2048,0.005596800148487091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,2048,0.01733013391494751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,2560,0.011239467064539592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,1536,0.004529066880544027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,1536,0.01726613243420919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,1536,0.009251200159390767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,1024,0.003605333218971888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,1024,0.014524799585342408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,2048,0.010554666320482891
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,768,0.0035445332527160645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,768,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,1024,0.008691199620564779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,512,0.003036800026893616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,512,0.014456533392270408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,768,0.007981866598129272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,256,0.0027072000006834666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,256,0.015516799688339234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,512,0.007897600034872691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,128,0.0025994665920734406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,128,0.014484266440073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,128,0.007097599903742473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,256,0.007562666634718577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,64,0.0022250667214393617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,64,0.014494933684666953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,4096,32,0.002501333256562551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,4096,32,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,65536,0.09063466389973959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,65536,0.08457173506418864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,16384,0.02951893409093221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,16384,0.03744959831237793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,5120,65536,0.1272703965504964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,12288,0.023180800676345825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,12288,0.02901653250058492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,12288,0.029768532514572142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,10240,0.01999680002530416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,10240,0.027030400435129803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,16384,0.03672959804534912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,8192,0.01568000018596649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,8192,0.02683626612027486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,8192,0.02185386617978414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,7168,0.013546666502952576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,7168,0.025076266129811602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,10240,0.02635519901911418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,6144,0.012357333302497863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,6144,0.02262826760609945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,7168,0.019834667444229126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,5120,0.010686933000882467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,5120,0.02055466572443644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,6144,0.018216532468795777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,4096,0.010423466563224792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,4096,0.01964906652768453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,4096,0.014170666535695394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,3584,0.009559466441472372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,5120,0.01648533344268799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,3584,0.01855573256810506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,3072,0.007134933272997539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,3072,0.018097066879272462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,4096,65536,0.12685866355895997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,2560,0.0066549330949783325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,2560,0.019078399737675986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,3072,0.012442666292190551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,2048,0.00520000010728836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,2048,0.01722773313522339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,3584,0.012888532876968384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,1536,0.0042357335488001505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,1536,0.01683626572291056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,1536,0.00962773362795512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,1024,0.0034122665723164878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,1024,0.014514133334159851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,2048,0.010409599542617798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,768,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,768,0.01532373329003652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,2560,0.011260799566904704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,512,0.003050666550795237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,512,0.017496534188588462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,1024,0.008728532989819845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,256,0.002752000093460083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,256,0.01446399986743927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,512,0.007896533111731212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,128,0.0026026666164398193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,128,0.015250133474667868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,768,0.008296533425649007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,64,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,64,0.01490239997704824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3584,32,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,128,0.007085866729418437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,256,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3584,32,0.01423466702302297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,65536,0.0730133295059204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,65536,0.07966399987538655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,16384,0.024985599517822265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,16384,0.03560853401819865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,12288,0.01954879959424337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,12288,0.03068160017331441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,12288,0.029316266377766926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,10240,0.017391999562581383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,10240,0.028702932596206664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,16384,0.03666666746139526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,8192,0.017145599921544394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,8192,0.026650667190551758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,10240,0.025158399343490602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,7168,0.015479466319084168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,7168,0.024757333596547446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,8192,0.02186346650123596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,6144,0.01344319979349772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,6144,0.024807467063268026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,7168,0.019471999009450278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,5120,0.011095466216405232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,5120,0.021102933088938396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,6144,0.01815999945004781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,4096,0.009465600053469341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,4096,0.021112533410390218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,5120,0.016160000363985697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,3584,0.008545066912968953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,3584,0.019471999009450278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,4096,0.014169599612553915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,3072,0.007607466479142506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,3072,0.019458132982254028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,3584,0.012890666723251343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,2560,0.00659093310435613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,2560,0.018281600872675576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,3072,0.012475732962290447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,2560,0.01121493379275004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,2048,0.005544533332188925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,2048,0.01700693368911743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3584,65536,0.12658027013142903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,1536,0.004678399860858917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,1536,0.01650773286819458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,1024,0.004322133461634318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,1024,0.0169322669506073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,1536,0.00955839951833089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,768,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,768,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,2048,0.010245333115259807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,512,0.0031167998909950255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,512,0.014612266421318054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,1024,0.008830933769543966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,256,0.00271573339899381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,256,0.014601600170135499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,768,0.008276266853014629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,512,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,128,0.0025973332424958544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,256,0.0075765331586201985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,128,0.014391466975212097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,64,0.0023359999060630797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,3072,32,0.0026719999810059865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,64,0.014605866869290671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,3072,32,0.01461120049158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,128,0.006742399930953979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,65536,0.06564586559931437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,16384,0.021926399072011313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,65536,0.07587200005849203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,16384,0.03397546609242757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,12288,0.017406932512919106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,12288,0.03062719901402791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,12288,0.029209599892298384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,10240,0.01580586632092794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,10240,0.027423999706904095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,16384,0.036685868104298906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,8192,0.015568000078201295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,8192,0.02531733314196269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,10240,0.025199999411900837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,7168,0.013392000397046407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,7168,0.023600000143051147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,8192,0.0213919997215271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,6144,0.012503467003504434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,6144,0.023181867599487305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,7168,0.01985493302345276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,5120,0.010627200206120808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,5120,0.02118826707204183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,6144,0.017842133839925133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,4096,0.009475200374921163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,4096,0.020336000124613444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,5120,0.016082132856051125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,3584,0.008242133259773254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,3584,0.019237333536148073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,4096,0.014138666788736978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,3072,0.007382399837176005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,3072,0.019170133272806804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,3584,0.013082666198412576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,2560,0.0067210664351781205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,2560,0.018369066715240478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,3072,0.012505599856376648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,2048,0.005929600199063619
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,2048,0.01769599914550781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,2560,0.011197866996129353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,1536,0.0050805335243542995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,1536,0.01844693422317505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,1536,0.00921066701412201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,1024,0.004212266703446706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,1024,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,2048,0.010428800185521444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,768,0.0037450666228930154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,768,0.015321600437164306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,1024,0.008740267157554627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,512,0.0034111998975276947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,768,0.008285866677761078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,512,0.015686399737993875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,256,0.002974933385848999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,256,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,256,0.007507200042406718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,128,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,512,0.0079434668024381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,128,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,64,0.002553600072860718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,64,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2560,32,0.002586666742960612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,128,0.0071263998746871945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2560,32,0.014434132973353067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,65536,0.055086934566497804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,65536,0.06853973070780436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,16384,0.01747093399365743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,16384,0.030989867448806763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,3072,65536,0.12689920266469318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,12288,0.01422826647758484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,12288,0.028907734155654907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,12288,0.02932586669921875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,10240,0.01285866697629293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,10240,0.02682346701622009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,16384,0.03660906553268432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,8192,0.012038399775822956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,8192,0.023783467213312783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,10240,0.02578666607538859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,7168,0.011404800415039062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,7168,0.02294293244679769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,8192,0.02150826652844747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,6144,0.010497066378593444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,6144,0.021093332767486574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,7168,0.019924267133076986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,5120,0.00997866690158844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,5120,0.019435733556747437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,6144,0.017809067169825235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,4096,0.00804799993832906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,4096,0.019538132349650066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,5120,0.016561067104339598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,3584,0.007484800120194752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,3584,0.018576000134150186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,4096,0.014152533809343972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,3072,0.006680533289909363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,3072,0.018172800540924072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,3584,0.013680000106493631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,2560,0.005340800185998281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,2560,0.017154133319854735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,3072,0.012033067146937053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,2048,0.0046517332394917805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,2048,0.016556800405184428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,2560,0.011617066462834676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,1536,0.00421973317861557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,1536,0.017029333114624023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,2048,0.010388267040252686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,1024,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,1536,0.009577600161234538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,1024,0.014593066771825156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,768,0.0033482665816942847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,768,0.015986133615175882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2560,65536,0.12653226852416993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,512,0.00296426663796107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,512,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,768,0.008429867029190064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,256,0.0027583998938401537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,1024,0.00879253347714742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,256,0.014522666732470194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,512,0.007551999886830647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,128,0.0022495999932289124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,128,0.014385066429773965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,64,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,64,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,2048,32,0.0022122666239738466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,256,0.007569066683451335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,128,0.00727040022611618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,65536,0.04199893474578857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,2048,32,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,16384,0.014227199554443359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,65536,0.060609066486358644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,16384,0.031167999903361006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,12288,0.012011733651161195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,12288,0.028429865837097168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,12288,0.028883200883865357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,10240,0.011704533298810323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,10240,0.026638933022816974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,16384,0.03625066677729289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,8192,0.012293333808581035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,8192,0.024664533138275147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,10240,0.02515733242034912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,7168,0.011008000373840332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,7168,0.022343466679255165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,8192,0.021652267376581828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,6144,0.009935999910036724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,6144,0.02092693249384562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,7168,0.019707733392715455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,5120,0.008551466464996337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,5120,0.021077332894007365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,6144,0.017864533265431724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,4096,0.0072543998559316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,4096,0.01981226603190104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,5120,0.01611199975013733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,3584,0.0066890666882197065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,3584,0.0185973326365153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,4096,0.01423893372217814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,3584,0.012827733159065246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,3072,0.0063391998410224915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,3072,0.01762666702270508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,2560,0.005522133409976959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,2560,0.01704960068066915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,2560,0.011252267162005107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,2048,0.005060266455014547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,2048,0.016749866803487144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,3072,0.012625066439310708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,1536,0.004373333354791006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,1536,0.01681386629740397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,1536,0.009613866607348125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,2048,0.010322133700052898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,1024,0.0034826666116714476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,1024,0.01486186683177948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,768,0.0030016000072161358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,768,0.015989333391189575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,768,0.008265600105126699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,512,0.002869333326816559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,1024,0.008777599533398945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,512,0.01590826710065206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,256,0.002570666621128718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,256,0.014378666877746582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,2048,65536,0.12658560276031494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,512,0.007522133489449819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,128,0.002313599983851115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,256,0.0075552001595497135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,128,0.014172800381978354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,64,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,64,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,128,0.007233066856861115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1536,32,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,65536,0.030308266480763752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1536,32,0.014658133188883463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,65536,0.053583999474843345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,16384,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,16384,0.02879253427187602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,12288,0.012102400263150532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,12288,0.02572373350461324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,12288,0.02882133324940999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,10240,0.010797866185506185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,10240,0.024753065903981526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,16384,0.03640426794687907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,8192,0.009799466530481974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,8192,0.02308373252550761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,10240,0.02509653369585673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,7168,0.00915839970111847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,7168,0.026636799176534016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,8192,0.02157013416290283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,6144,0.00846613347530365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,6144,0.023893332481384276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,7168,0.019361066818237304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,5120,0.0079434668024381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,5120,0.022788266340891518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,6144,0.017968000968297322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,4096,0.00716480016708374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,4096,0.021359999974568687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,5120,0.01609386702378591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,3584,0.010744532942771912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,3584,0.01809920072555542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,4096,0.014166399836540222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,3072,0.00965226689974467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,3072,0.018103466431299845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,3584,0.012878933548927307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,2560,0.008570667107899983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,2560,0.018696532646814982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,3072,0.012502400080362954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,2048,0.007333333293596904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,2048,0.017025067408879598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,2560,0.011185066898663838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,1536,0.004244266450405121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,1536,0.01681493322054545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,2048,0.010501333077748616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,1024,0.0035413332283496858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,1024,0.017138133446375527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,1536,0.009147733449935913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,768,0.0033802665770053864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,768,0.01564479966958364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,1024,0.008455466230710347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,512,0.003048533449570338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,768,0.008307200173536937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,512,0.016800000270207723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,256,0.002682666728893916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,256,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,512,0.007536000013351441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,128,0.002436266591151555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,128,0.014386133352915446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,256,0.007624533275763195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,128,0.007145600020885467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,64,0.002282666663328807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,64,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,1024,32,0.0022474666436513263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,1024,32,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,65536,0.026294400294621784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,65536,0.05155733426411947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,16384,0.011751466989517212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,16384,0.028629332780838013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1536,65536,0.12623466650644938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,12288,0.012148267030715943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,12288,0.025434666872024538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,12288,0.02900693416595459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,10240,0.01109333336353302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,10240,0.024800000588099162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,16384,0.03583253224690755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,8192,0.00960213343302409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,8192,0.02321386734644572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,8192,0.02109439969062805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,7168,0.008820266524950663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,7168,0.02071253259976705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,10240,0.02565760016441345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,6144,0.008147199948628742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,6144,0.021322667598724365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,7168,0.019438934326171876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,5120,0.007574399809042613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,5120,0.02023893396059672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,6144,0.01830293337504069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,4096,0.006814933319886525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,4096,0.019270400206247963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,5120,0.016158933440844216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,3584,0.00629013329744339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,3584,0.018037333091100057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,4096,0.01421440045038859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,3072,0.005618133147557576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,1024,65536,0.12581439812978107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,3072,0.018963199853897095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,3584,0.013218133648236593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,2560,0.005061333378156027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,2560,0.01656000018119812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,3072,0.01211199959119161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,2048,0.004729599754015604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,2048,0.016657066345214844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,2560,0.011174399654070537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,1536,0.00382080003619194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,1536,0.016219733158747356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,1024,0.00388373335202535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,2048,0.010411733388900756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,1024,0.016017066438992818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,768,0.003257599969704946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,1536,0.009592533111572266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,768,0.014733866850535075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,512,0.002898133297761281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,1024,0.008403199911117553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,512,0.014723199605941772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,256,0.0027583998938401537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,768,0.007905066510041555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,256,0.01467626690864563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,512,0.007573333382606506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,128,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,256,0.00713919997215271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,128,0.014541866381963095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,64,0.0021717332303524016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,768,32,0.002181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,128,0.007192533214886982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,64,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,65536,0.018914133310317993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,768,32,0.01444586714108785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,16384,0.011226666967074077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,65536,0.04655040105183919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,16384,0.027589333057403565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,12288,0.009538132945696514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,12288,0.02476159930229187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,12288,0.028535467386245728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,10240,0.008775466680526733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,10240,0.022977066040039063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,16384,0.03602453470230103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,8192,0.007995733122030894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,8192,0.02295893430709839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,10240,0.025195733706156416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,7168,0.007512533167997997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,7168,0.022709333896636964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,8192,0.02165973385175069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,6144,0.006971733272075653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,6144,0.020994132757186888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,7168,0.019778132438659668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,5120,0.006603733201821645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,5120,0.0198634664217631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,6144,0.018231467405954997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,4096,0.00639573335647583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,4096,0.018757333358128868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,5120,0.01630400021870931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,3584,0.006046933432420095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,3584,0.018024533987045288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,4096,0.014176000157992044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,3072,0.0059797331690788266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,3584,0.012889599800109864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,3072,0.01756160060564677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,2560,0.005463466544946035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,768,65536,0.12305173079172771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,2560,0.017223467429478966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,2048,0.004755199948946635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,2048,0.017918932437896728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,3072,0.011988266309102377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,1536,0.0038399999340375268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,2560,0.011374933520952861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,1536,0.016370133558909098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,1024,0.00327360009153684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,2048,0.010437333583831787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,1024,0.01513813336690267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,768,0.0030720000465710956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,768,0.015222400426864624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,1024,0.008316799998283386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,1536,0.009705600142478944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,512,0.002661333233118057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,768,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,512,0.016288000345230102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,256,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,256,0.014259200294812521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,128,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,512,0.007621333499749501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,256,0.007202133536338806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,64,0.002169599880774816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,128,0.01383680005868276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,512,32,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,128,0.007121066749095917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,64,0.014491732915242514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,512,32,0.014522666732470194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,65536,0.016976000865300496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,65536,0.041110400358835855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,16384,0.008403199911117553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,16384,0.02916373411814372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,12288,0.007096533477306366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,12288,0.023915733893712363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,12288,0.028463999430338543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,16384,0.03586879968643188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,10240,0.0065610667069753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,10240,0.02447999914487203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,10240,0.02515946626663208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,8192,0.00874133308728536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,8192,0.023105067014694215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,7168,0.008407466610272725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,7168,0.020986666282018028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,8192,0.021568000316619873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,6144,0.006366933385531108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,7168,0.019801600774129232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,6144,0.021862399578094483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,5120,0.006794666747252147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,5120,0.020838399728139244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,6144,0.017779199282328288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,4096,0.005873066683610281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,5120,0.016280532876650492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,4096,0.01850559910138448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,3584,0.005635199944178263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,3584,0.01707093318303426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,4096,0.013742933670679728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,3584,0.013176533579826354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,3072,0.005621333420276642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,3072,0.01770133376121521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,2560,0.005100800096988678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,512,65536,0.12322773138682049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,2560,0.016482133666674295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,2048,0.004276266694068909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,2048,0.017553067207336424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,3072,0.012113066514333089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,1536,0.003866666555404663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,2560,0.011668266852696736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,1536,0.016749866803487144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,1024,0.003443199892838796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,2048,0.009961600104967754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,1024,0.016446933150291443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,768,0.0030570665995279947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,1536,0.009644800424575805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,768,0.016167466839154564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,1024,0.008353066444396973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,512,0.002666666607062022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,512,0.015809067090352378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,256,0.0022250667214393617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,768,0.008397866288820903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,512,0.007612800101439159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,256,0.014204800128936768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,128,0.0023018665611743926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,256,0.007228800157705943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,128,0.013421866297721862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,64,0.0022624000906944276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,256,32,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,128,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,64,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,65536,0.015030399958292643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,256,32,0.0132533331712087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,65536,0.03713599840799968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,16384,0.008827733000119527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,16384,0.024885332584381102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,12288,0.00772159993648529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,12288,0.022702932357788086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,256,65536,0.12284266948699951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,10240,0.006054399907588959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,10240,0.023142399390538533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,12288,0.028460800647735596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,16384,0.036202665170033774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,8192,0.0058783998092015585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,8192,0.022056533892949422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,7168,0.005523199836413065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,7168,0.020829866329828896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,10240,0.025212800502777098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,6144,0.006252799928188324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,8192,0.021025067567825316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,6144,0.02135146657625834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,5120,0.005946666498978933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,7168,0.01989333430926005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,5120,0.020294400056203206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,4096,0.005543466905752818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,4096,0.019474132855733236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,6144,0.017806933323542277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,3584,0.0052159999807675685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,3584,0.01764586567878723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,5120,0.016260266304016113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,3072,0.005110399921735128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,4096,0.013667200009028116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,3072,0.01718719998995463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,2560,0.005498666564623515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,3584,0.012834133704503379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,2560,0.016739199558893837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,2048,0.00514026681582133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,3072,0.012410666545232136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,2048,0.016778665781021117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,1536,0.0038261334101359046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,2560,0.01123413344224294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,1536,0.015867732961972556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,1024,0.0030570665995279947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,2048,0.010036266843477885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,1024,0.01474346617857615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,1536,0.00918826659520467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,768,0.003010133405526479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,1024,0.008335999647776286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,768,0.015899733702341715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,512,0.0026677332818508146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,768,0.00801386684179306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,512,0.014521599809328715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,256,0.002179199953873952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,256,0.013832533359527588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,512,0.007982933521270752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,128,0.00225600004196167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,256,0.007068799932797749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,128,0.0145087997118632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,128,0.006762666503588359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,64,0.002217599997917811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,64,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,128,32,0.0022442666192849475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,65536,0.0141184002161026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,128,32,0.014259200294812521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,16384,0.007115733126799266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,65536,0.03652799924214681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,16384,0.024934399127960204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,12288,0.006343466540177663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,160,128,65536,0.12206506729125977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,12288,0.02304960091908773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,10240,0.006377600133419037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,8192,0.00588266650835673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,7168,0.005448533097902933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,10240,0.022768000761667885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,8192,0.0207370658715566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,6144,0.00558186670144399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,7168,0.020760534207026164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,5120,0.005269333223501841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,6144,0.02095680038134257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,4096,0.0050901333491007485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,5120,0.02038933237393697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,3584,0.005547733108202616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,4096,0.019156267245610557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,3584,0.018105600277582803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,3072,0.005241600175698599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,2560,0.005096533397833506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,3072,0.017248000701268515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,2048,0.0050805335243542995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,1536,0.0037845333417256674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,2560,0.01713706652323405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,2048,0.017494400342305504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,1024,0.003303466737270355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,1536,0.01607253352801005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,768,0.0031317333380381264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,1024,0.01602453291416168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,512,0.002775466690460841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,768,0.014780799547831217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,256,0.0023391999304294584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,512,0.014670933286348978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,128,0.0023658665517965953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,256,0.01444586714108785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,64,0.002195200075705846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,64,32,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,128,0.0144405335187912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,64,0.014442666371663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,65536,0.009947733084360758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,64,32,0.01452906628449758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,16384,0.006704000135262807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,65536,0.03499840100606282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,12288,0.005470933516820272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,10240,0.0058783998092015585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,12288,0.022802132368087768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,16384,0.024851200977961223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,8192,0.00550186683734258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,10240,0.021780266364415487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,7168,0.005756799876689911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,8192,0.021921066443125407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,6144,0.0051360001166661584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,7168,0.020729599396387736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,5120,0.005063466727733612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,4096,0.0046304002404212955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,5120,0.020499199628829956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,6144,0.021166932582855225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,3584,0.005092266698678335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,4096,0.01935466726620992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,3072,0.005184000233809153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,2560,0.005031466484069824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,3584,0.017233065764109292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,2048,0.004673066735267639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,3072,0.017604267597198485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,1536,0.003822933385769526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,2560,0.016683733463287352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,2048,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,1024,0.003066666672627131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,768,0.0030026666820049284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,1536,0.015468800067901611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,1024,0.014572800199190775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,512,0.0026015999416510267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,768,0.014682666460673014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,256,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,128,0.0022026665508747103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,512,0.014908799529075622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,256,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,64,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,128,0.014350933829943338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,160,32,32,0.0022240000466505687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,32,0.01437440017859141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,160,32,64,0.01458560029665629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,16384,0.21416212717692057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,16384,0.3761290550231934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,12288,0.1638005256652832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,12288,0.27771307627360026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,10240,0.23881492614746094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,10240,0.1417781352996826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,8192,0.197325865427653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,8192,0.12021546363830567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,8192,0.06264106829961141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,7168,0.164356263478597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,7168,0.10192746321360271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,10240,0.07317439715067545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,6144,0.14605120023091633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,6144,0.09080533186594644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,12288,0.08601493040720622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,5120,0.11974613666534424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,5120,0.07878399689992269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,6144,0.047747198740641275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,4096,0.09865600268046061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,4096,0.06662613153457642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,7168,0.054523734251658115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,3584,0.08564693133036295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,3584,0.06008106470108032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,5120,0.04179626703262329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,3072,0.07427199681599936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,3072,0.053471998373667395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,16384,0.11304960250854493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,2560,0.06348479986190796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,2560,0.047386666138966874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,3584,0.03140693306922913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,2048,0.053157333532969156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,2048,0.042827733357747394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,4096,0.03467839956283569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,1536,0.03971199989318848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,1536,0.035019731521606444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,3072,0.028445865710576373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,1024,0.027497599522272747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,1024,0.027479465802510577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,2560,0.025209599733352663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,768,0.021642667055130006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,768,0.02699413299560547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,2048,0.021541333198547362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,512,0.01493119994799296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,512,0.02145919998486837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,1536,0.0179584006468455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,256,0.009806933005650838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,256,0.01863573392232259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,256,0.009205333391825358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,128,0.006618666648864746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,128,0.018653867642084758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,1024,0.013893333077430726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,64,0.006307200094064077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,64,0.01864746610323588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,65536,32,0.008377599716186523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,65536,32,0.020668800671895346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,65536,0.31532586415608727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,65536,0.24760212898254394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,128,0.009147733449935913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,16384,0.09199146429697672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,16384,0.07317226727803548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,512,0.010590933760007222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,12288,0.06294933160146078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,12288,0.0591264009475708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,65536,768,0.01251520017782847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,10240,0.05397760073343912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,10240,0.05186239878336588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,10240,0.029315199454625445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,8192,0.04527573188145955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,8192,0.04442773262659709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,12288,0.03320106665293376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,7168,0.038439468542734785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,7168,0.0400000015894572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,16384,0.042580266793568924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,6144,0.03411733309427897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,6144,0.03488746484120687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,8192,0.02456106742223104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,5120,0.02940053343772888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,5120,0.03147839903831482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,7168,0.021971199909845987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,4096,0.02442880074183146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,4096,0.02837439974149068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,5120,0.017935999234517417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,3584,0.021710934241612752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,3584,0.026999467611312868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,6144,0.019814399878184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,3072,0.01884373426437378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,3072,0.023612799247105916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,4096,0.015131733814875283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,2560,0.016081066926320393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,2560,0.022695465882619222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,2560,0.01204266647497813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,2048,0.013250133395195008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,2048,0.02061333258946737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,3584,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,1536,0.010706133643786113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,1536,0.020499199628829956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,3072,0.01332586705684662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,1024,0.007977599898974102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,1024,0.017102932929992674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,1024,0.009090133508046468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,768,0.006995200117429097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,768,0.018526933590571084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,1536,0.009964799880981446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,512,0.004677333434422811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,512,0.016009599963823954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,2048,0.010844799876213073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,256,0.0034400001168251038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,256,0.015388799707094827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,512,0.008007466793060303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,128,0.00347626656293869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,128,0.014985600113868713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,768,0.00851093331972758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,64,0.00310506671667099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,64,0.014640000462532044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,16384,32,0.0030634666482607523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,16384,32,0.014999467134475707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,128,0.007495466868082683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,65536,0.23752320607503258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,65536,0.18805759747823078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,16384,0.06724906762440999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,256,0.007207466661930085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,16384,0.05904959837595621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,12288,0.0482101321220398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,12288,0.04911893208821615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,12288,0.0313045342763265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,10240,0.04136213461558024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,10240,0.043409065405527754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,16384,0.039614931742350264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,8192,0.033838931719462076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,8192,0.03726613521575928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,10240,0.027051732937494917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,7168,0.02967360019683838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,7168,0.032782934109369916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,8192,0.022725333770116173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,6144,0.02654079993565877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,6144,0.029241599639256793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,7168,0.020733867088953653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,5120,0.022600533564885457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,5120,0.02704426646232605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,6144,0.0186463991800944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,4096,0.01851200064023336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,4096,0.02459519902865092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,5120,0.017443199952443443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,3584,0.016696532567342125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,3584,0.02347946763038635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,4096,0.01493119994799296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,3072,0.014546133081118264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,3072,0.022278400262196858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,3584,0.01411733329296112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,2560,0.012566399574279786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,2560,0.02250986695289612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,3072,0.012863999605178833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,2048,0.010564266641934713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,2048,0.019704532623291016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,2560,0.011699199676513672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,1536,0.008845866719881693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,1536,0.018702934185663857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,1536,0.009604266285896302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,1024,0.007322666545708974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,1024,0.017222400506337485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,2048,0.010471466183662414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,768,0.005083733300367991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,768,0.01750826636950175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,768,0.008303999900817871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,512,0.004203733305136363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,512,0.014694399634997048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,1024,0.008533333738644917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,256,0.003566933423280716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,256,0.016505600015322367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,512,0.007919999957084655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,128,0.003421866645415624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,256,0.0072618668278058365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,128,0.014944000045458474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,64,0.0030432000756263735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,64,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,12288,32,0.003472000112136205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,12288,32,0.014722133676211039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,65536,0.19647253354390462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,65536,0.16700372695922852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,128,0.007319466769695282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,16384,0.05643413464228312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,16384,0.052707199255625406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,16384,0.03836053212483724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,12288,0.04078293244043986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,12288,0.043323731422424315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,12288,0.03048959970474243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,10240,0.03456213474273682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,10240,0.039287467797597245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,16384,65536,0.14613439242045084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,8192,0.02900693416595459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,8192,0.03351680040359497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,10240,0.025993599494298296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,7168,0.025272534290949507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,7168,0.02987839976946513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,8192,0.02186346650123596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,6144,0.02260266741116842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,6144,0.028407466411590577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,7168,0.020243199666341145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,5120,0.019511467218399046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,5120,0.025895466407140095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,12288,65536,0.13341333071390787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,4096,0.015428266922632852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,4096,0.02300373315811157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,6144,0.018633600076039633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,3584,0.0137855996688207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,3584,0.022851200898488362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,4096,0.014518400033315023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,3072,0.012177067001660664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,3072,0.020971733331680297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,5120,0.017012266318003337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,2560,0.01083626647790273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,2560,0.02135466734568278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,3584,0.013782399892807006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,2048,0.009136000275611877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,2048,0.020097066958745323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,2560,0.011283199985822041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,1536,0.007559466858704884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,1536,0.01731520096460978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,3072,0.012456533312797547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,1024,0.005471999943256378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,1024,0.016472533345222473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,1024,0.008749866485595703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,768,0.004229333500067393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,768,0.014625066518783569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,1536,0.009585066636403402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,512,0.003869866579771042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,512,0.015448533495267234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,768,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,256,0.003403733422358831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,256,0.014937600493431092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,512,0.007956266899903615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,128,0.00310506671667099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,128,0.015561599532763162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,256,0.007506133119265239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,64,0.0030559999247392017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,128,0.007143466671307881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,10240,32,0.003044266750415166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,64,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,10240,32,0.015401599804560342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,2048,0.009995733698209126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,65536,0.18600106239318848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,65536,0.1281162659327189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,16384,0.049899733066558837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,16384,0.04760426680246989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,12288,0.03672106663386027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,12288,0.03914026816685994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,12288,0.029705599943796797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,10240,0.03120959997177124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,10240,0.03501760164896647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,16384,0.03752533197402954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,8192,0.026494934161504106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,8192,0.029356799523035687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,10240,0.025597866376241046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,7168,0.022638932863871256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,7168,0.02709226608276367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,8192,0.022083199024200438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,6144,0.02041920026143392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,6144,0.025489066044489545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,7168,0.020337067047754922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,5120,0.01722986698150635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,5120,0.024489599466323852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,10240,65536,0.12874133586883546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,4096,0.014011733730634055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,4096,0.02204373280207316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,6144,0.01880106727282206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,3584,0.011766399939854939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,3584,0.02078826626141866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,5120,0.016743467251459757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,3072,0.010602666934331258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,3072,0.020909865697224937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,4096,0.014456533392270408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,2560,0.009146666526794434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,2560,0.019423999389012656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,3584,0.013369599978129068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,2048,0.00804799993832906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,2048,0.016839466492335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,2560,0.01125333309173584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,1536,0.0066431999206542965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,1536,0.01717653274536133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,2048,0.010219732920328777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,1024,0.004816000163555145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,1024,0.016703999042510985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,3072,0.012202666203180949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,768,0.004309333364168803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,768,0.01655359963575999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,768,0.008086400230725606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,512,0.0035466666022936503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,512,0.01537493367989858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,1024,0.008809600273768108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,256,0.003047466774781545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,256,0.014622933665911355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,1536,0.009159466624259949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,128,0.003033600002527237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,256,0.00719893326361974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,128,0.014467199643452963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,64,0.0028533334533373516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,64,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,8192,32,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,8192,32,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,512,0.00763626645008723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,128,0.0071839998165766404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,65536,0.1607583999633789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,65536,0.13280746936798096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,16384,0.04078400135040283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,16384,0.041433600584665935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,12288,0.030096000432968138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,12288,0.03562026818593343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,12288,0.029208532969156903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,10240,0.02610879937807719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,10240,0.031192533175150555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,16384,0.03751039902369181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,8192,0.02156053384145101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,8192,0.026846933364868163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,10240,0.025672533114751178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,7168,0.01881706714630127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,7168,0.025783467292785644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,8192,0.022098133961359658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,6144,0.01672533353169759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,6144,0.02547946572303772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,7168,0.02028053402900696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,5120,0.014241066575050355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,5120,0.022822399934132896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,6144,0.01835626761118571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,4096,0.011497599879900615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,4096,0.021306665738423665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,5120,0.016849066813786825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,3584,0.010381866494814556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,3584,0.021359999974568687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,4096,0.014176000157992044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,3072,0.009521067142486572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,3072,0.018900267283121743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,3584,0.013433600465456644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,2560,0.00837546686331431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,2560,0.018279467026392618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,3072,0.012054399649302164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,2048,0.007169066866238911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,2048,0.01655466655890147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,2560,0.01118933359781901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,1536,0.005924266576766968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,1536,0.01648640036582947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,2048,0.009995733698209126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,1024,0.004023466755946477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,1024,0.016453333695729575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,1536,0.00962773362795512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,768,0.003565866748491923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,768,0.01668693423271179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,1024,0.008748799562454224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,512,0.003078400095303853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,512,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,768,0.00798826664686203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,512,0.007593599955240886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,256,0.002963199963172277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,256,0.014573867122332254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,128,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,128,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,128,0.007142400244871776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,64,0.0026208000878492994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,64,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,7168,32,0.0026229334374268847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,7168,32,0.014640000462532044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,256,0.007380266487598419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,65536,0.14506346384684246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,65536,0.10546133518218995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,16384,0.03650453488032023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,16384,0.039469865957895915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,8192,65536,0.1274506648381551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,12288,0.026997333765029906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,12288,0.03298880060513814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,16384,0.036697598298390706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,10240,0.02367786765098572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,10240,0.030562132596969604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,12288,0.028837333122889202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,8192,0.01943040092786153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,8192,0.02691733241081238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,10240,0.025641600290934246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,7168,0.0170197327931722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,7168,0.024868265787760416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,8192,0.02186346650123596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,6144,0.0151936004559199
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,6144,0.023574399948120116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,7168,0.020204800367355346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,5120,0.013138133287429809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,5120,0.022690133253733317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,6144,0.018179200092951455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,4096,0.010770133137702942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,4096,0.020693333943684895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,5120,0.016609066724777223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,3584,0.009970133503278095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,3584,0.02053119937578837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,4096,0.01368106702963511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,3072,0.008748799562454224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,3072,0.01883946657180786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,3584,0.013336533308029174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,2560,0.007930666704972585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,2560,0.018106667200724284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,3072,0.012105600039164225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,2048,0.006704000135262807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,2048,0.016569599509239197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,2560,0.011251200238863628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,1536,0.005374933282534281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,2048,0.010056533416112264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,1536,0.017196800311406454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,7168,65536,0.1263935963312785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,1024,0.0038912000755469004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,1024,0.01584106683731079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,768,0.0040554667512575785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,768,0.016717867056528727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,768,0.008412800232569377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,512,0.003676799933115641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,512,0.01469546655813853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,1024,0.008309333523114523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,256,0.0029898665845394133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,256,0.015065600474675497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,256,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,1536,0.009698133667310078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,128,0.0029109333952267963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,128,0.014536533753077188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,64,0.002746666719516118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,64,0.014673067132631936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,6144,32,0.002962133288383484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,512,0.00757120003302892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,128,0.007083733379840851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,6144,32,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,65536,0.11363306840260823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,65536,0.09445013205210367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,16384,0.03683306773503621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,16384,0.042947200934092204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,12288,0.02940906683603923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,12288,0.02997973362604777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,12288,0.02882133324940999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,10240,0.021834667523701987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,10240,0.028974932432174683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,16384,0.036347734928131106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,8192,0.017811199029286705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,8192,0.025329067309697466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,10240,0.025595732529958087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,7168,0.015247999628384908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,7168,0.024678399165471397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,8192,0.02190720041592916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,6144,0.013711999853452048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,6144,0.022936532894770302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,7168,0.02023893396059672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,5120,0.012018133203188579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,5120,0.02169493238131205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,6144,0.018292266130447387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,4096,0.009872000416119893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,4096,0.020680532852808634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,5120,0.016133333245913185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,3584,0.008946133653322856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,3584,0.018679465850194296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,4096,0.014137599865595499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,3072,0.008381866415341695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,3072,0.019394133488337198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,3584,0.012922666470209756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,2560,0.007334400216738384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,2560,0.016744534174601235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,3072,0.012388267119725545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,2048,0.006423466900984447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,2560,0.010995200276374817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,2048,0.01684160033861796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,1536,0.005142400165398916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,1536,0.016829866170883178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,6144,65536,0.1264469305674235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,1024,0.003962666789690653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,1024,0.015582933028539022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,1536,0.009629866480827332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,768,0.0038143999874591826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,768,0.015296000242233276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,1024,0.008814932902654012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,512,0.003303466737270355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,512,0.014897066354751586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,2048,0.010028800368309021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,256,0.0030080000559488933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,256,0.014811733365058899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,768,0.00798933357000351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,128,0.002739199995994568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,128,0.014702933033307395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,256,0.0071487997968991595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,64,0.0026079999903837843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,64,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,5120,32,0.002593066543340683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,5120,32,0.014502400159835815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,512,0.008020266890525818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,128,0.007190399865309398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,65536,0.0992469310760498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,65536,0.08584213256835938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,16384,0.030065067609151203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,16384,0.03622506856918335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,12288,0.024360533555348715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,12288,0.028970666726430255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,12288,0.029254400730133058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,10240,0.018436266978581747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,10240,0.026819199323654175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,16384,0.036177066961924235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,8192,0.014985600113868713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,8192,0.024495999018351235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,10240,0.025663999716440837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,7168,0.013390933473904928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,7168,0.022797866662343343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,8192,0.021563732624053956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,6144,0.01218666632970174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,6144,0.022588799397150673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,7168,0.019825067122777304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,5120,0.0106495996316274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,5120,0.020834134022394816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,6144,0.017756799856821694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,4096,0.008717866738637288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,4096,0.01972480018933614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,5120,0.016292267044385276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,3584,0.008056533336639405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,3584,0.01926400065422058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,4096,0.0141184002161026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,3072,0.00719893326361974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,3072,0.018487467368443807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,3584,0.012974933783213297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,2560,0.0064298664530118305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,2560,0.016775466998418174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,3072,0.012013866504033407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,2048,0.005470933516820272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,2048,0.016901334126790367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,2560,0.011354666948318482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,1536,0.004332800209522247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,1536,0.016739199558893837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,2048,0.010334933797518413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,1024,0.003433600068092346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,1024,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,1536,0.009290666381518046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,768,0.0032042667269706728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,768,0.01535040040810903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,1024,0.00902186632156372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,768,0.008045866588751475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,512,0.003020799905061722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,512,0.015558399756749473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,256,0.002669866631428401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,256,0.014564266800880432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,256,0.007625600198904674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,512,0.007513600091139476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,128,0.002752000093460083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,128,0.015056000153223673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,64,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,64,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,4096,32,0.0029781334102153777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,4096,32,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,128,0.007176533341407776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,65536,0.08781119982401529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,65536,0.07867840131123861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,16384,0.025294933716456098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,16384,0.03410559892654419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,5120,65536,0.12610560258229572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,12288,0.020422399044036865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,12288,0.02723840077718099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,16384,0.03635093371073405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,10240,0.017869865894317626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,10240,0.025990400711695356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,12288,0.029174399375915528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,8192,0.01411946713924408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,8192,0.024939733743667602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,10240,0.02555733323097229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,8192,0.02133013407389323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,7168,0.012536533673604331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,7168,0.022668800751368203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,6144,0.011347200473149617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,6144,0.02093013326327006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,6144,0.017372800906499227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,5120,0.010057600339253743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,5120,0.0202293336391449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,7168,0.019194666544596353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,4096,0.008368000388145447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,4096,0.01872426668802897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,5120,0.015727999806404113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,3584,0.007786666850248973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,4096,0.013710932930310568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,3584,0.01918720006942749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,3072,0.0068917334079742435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,3072,0.018279467026392618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,4096,65536,0.12560959657033283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,2560,0.0059797331690788266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,3072,0.012030933300654094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,2560,0.017785600821177163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,2048,0.005106133222579956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,2048,0.0182805339495341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,3584,0.012917332847913108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,1536,0.004232533276081085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,1536,0.017018665870030723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,1536,0.009193600217501322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,1024,0.003502933432658514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,1024,0.01583146651585897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,2048,0.010362666845321656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,768,0.003454933315515518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,768,0.017072000106175742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,2560,0.011204266548156738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,512,0.003005866706371307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,512,0.014465066790580749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,1024,0.008721066514650981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,256,0.0026496000587940215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,256,0.014447999993960061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,512,0.007889066636562348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,768,0.007972266773382823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,128,0.0025568000972270967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,128,0.015326933066050211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,128,0.007066666583220164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,256,0.0071487997968991595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,64,0.002570666621128718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3584,32,0.0029845332105954488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,64,0.014698666334152222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3584,32,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,65536,0.07137813568115234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,65536,0.07200427055358886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,16384,0.024027733008066814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,16384,0.03302506605784099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,12288,0.019179733594258626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,12288,0.02995733420054118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,12288,0.028862933317820232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,10240,0.01646719972292582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,10240,0.02654293378194173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,16384,0.036235733826955156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,8192,0.013529599706331889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,8192,0.023036799828211465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,10240,0.025235199928283693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,7168,0.012079999844233195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,7168,0.02392959992090861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,8192,0.021035732825597127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,6144,0.010939733187357584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,6144,0.02216426730155945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,7168,0.019256534179051717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,5120,0.00962559978167216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,5120,0.02026346723238627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,6144,0.017777067422866822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,4096,0.007959466675917308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,4096,0.01888213356335958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,5120,0.01565439999103546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,3584,0.007344000041484833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,3584,0.018554667631785073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,4096,0.013658666610717773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,3072,0.006541866560777028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,3072,0.017809067169825235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,3072,0.012016000350316365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,3584,0.01293333371480306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,2560,0.005762133498986562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,2560,0.018870399395624796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,2048,0.005250133574008942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,2048,0.01807253360748291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3584,65536,0.1256223996480306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,1536,0.0042346666256586705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,1536,0.016645333170890807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,1536,0.009215999643007915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,1024,0.0037941334148248037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,1024,0.014607999722162882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,2048,0.010396800438563029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,768,0.003610666592915853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,768,0.016705065965652466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,2560,0.011281067132949829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,512,0.0031776001056035364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,512,0.014594133694966635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,512,0.007561600208282471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,256,0.0028917332490285236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,768,0.008360532919565837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,1024,0.008344533046086629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,128,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,256,0.0160671999057134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,128,0.014873600006103516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,64,0.002583466718594233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,64,0.014524799585342408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,3072,32,0.0030069333811601003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,128,0.007225599884986877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,3072,32,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,256,0.007187200089295705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,65536,0.06388373374938965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,16384,0.02153173287709554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,65536,0.06776533126831055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,16384,0.03214720090230306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,12288,0.017118932803471883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,12288,0.02863466739654541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,12288,0.028871466716130573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,10240,0.0153546671072642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,10240,0.024925865729649863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,16384,0.03589013417561849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,8192,0.01251200040181478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,8192,0.02265920042991638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,10240,0.025116799275080363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,7168,0.012097066640853882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,7168,0.021987199783325195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,8192,0.020716800292332967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,6144,0.010994133353233338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,6144,0.02140799959500631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,7168,0.01914773384730021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,5120,0.009482666850090027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,5120,0.019101866086324058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,6144,0.01744106610616048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,4096,0.007625600198904674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,4096,0.01955733299255371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,5120,0.015962666273117064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,3584,0.006808533271153767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,3584,0.019115734100341796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,4096,0.013677866260210673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,3072,0.005904000004132589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,3072,0.01686720053354899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,3584,0.013387733697891235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,2560,0.005538133283456167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,3072,0.011667199929555257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,2560,0.017122133572896322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,2048,0.004852266609668731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,2048,0.016822399695714314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,2560,0.01127893328666687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,1536,0.004284800092379252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,1536,0.017128533124923705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,2048,0.009995733698209126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,1024,0.0036746665835380556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,1024,0.015434666474660238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,1536,0.009318400422732036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,768,0.003502933432658514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,768,0.015205333630243937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,768,0.008006399869918824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,1024,0.008342400193214417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,512,0.003125333289305369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,512,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,256,0.0025759999950726825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,256,0.014877866705258688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,256,0.007055999835332234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,128,0.0025205334027608235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,128,0.015362133582433065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,512,0.00759680022795995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,64,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,3072,65536,0.12587733268737794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2560,32,0.0023413332800070446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,64,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,128,0.006838400165239971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,65536,0.055669331550598146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2560,32,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,16384,0.016707199811935424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,65536,0.06030186812082926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,16384,0.029228800535202028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,12288,0.012965333461761475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,12288,0.027831466992696126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,12288,0.02847893238067627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,10240,0.01250986655553182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,10240,0.024871466557184856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,16384,0.035846400260925296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,8192,0.012292266885439555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,8192,0.022716800371805825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,10240,0.024539732933044435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,7168,0.011321600278218586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,7168,0.020890667041142782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,8192,0.02068586746851603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,6144,0.009944533308347065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,6144,0.02067199945449829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,7168,0.019529600938161217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,5120,0.00888320008913676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,5120,0.019607466459274293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,6144,0.01773866613705953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,4096,0.007371733089288075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,4096,0.018565332889556883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,5120,0.01591253379980723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,3584,0.006828799843788147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,4096,0.01366933286190033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,3584,0.017579734325408936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,3072,0.0058218667904535925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,3072,0.01894186735153198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,3584,0.012853333353996277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,3072,0.01216319998105367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,2560,0.005016533533732096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,2560,0.01731733282407125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,2048,0.004748799900213877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,2048,0.016594133774439492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,2560,0.011251200238863628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,1536,0.003908266623814901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,1536,0.016886399189631144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,2048,0.01051093339920044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,1024,0.0034688000877698263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,1024,0.016427733500798545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2560,65536,0.12578346729278564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,768,0.003159466634194056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,1024,0.00876693328221639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,1536,0.009589333335558574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,768,0.015710933009783427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,512,0.002632533262173335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,512,0.014601600170135499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,256,0.0026122666895389557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,256,0.015979733069737753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,512,0.0076341331005096436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,128,0.002170666555563609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,256,0.0073728000124295544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,768,0.008327466746171314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,128,0.015588266650835672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,64,0.002242133269707362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,2048,32,0.002204799900452296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,64,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,65536,0.04087040026982625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,128,0.006929066777229309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,2048,32,0.013928533593813578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,16384,0.014148267110188803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,65536,0.05374826590220133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,16384,0.02760106722513835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,12288,0.011597866813341778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,12288,0.02609279950459798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,12288,0.028574933608373005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,10240,0.01070186694463094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,10240,0.025016532341639204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,16384,0.035520001252492266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,8192,0.009522133072217305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,8192,0.022747733195622764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,10240,0.025141332546869916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,7168,0.008995200196901958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,7168,0.02114026745160421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,8192,0.020994132757186888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,6144,0.007750399907430013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,6144,0.020501333475112914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,7168,0.019598933060963948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,5120,0.008589866757392883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,5120,0.019887999693552653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,6144,0.01739733417828878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,4096,0.007097599903742473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,4096,0.01917333404223124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,5120,0.015846400459607442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,3584,0.006357333560784657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,3584,0.018474666277567546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,4096,0.013306666413942972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,3072,0.005457066496213277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,3072,0.016551466782887776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,3584,0.013102933764457703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,2560,0.005162666738033295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,2560,0.016702934106191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,3072,0.011990400155385335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,2048,0.004574933151404063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,2048,0.01755626598993937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,2560,0.011334400375684102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,1536,0.003946666667858759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,1536,0.017845332622528076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,2048,0.010368000467618306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,1024,0.00344106654326121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,1024,0.014825600385665893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,1536,0.009352533022562663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,768,0.0031466667850812277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,768,0.01461013356844584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,1024,0.008307200173536937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,512,0.00308693324526151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,512,0.014443733294804893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,768,0.008481066425641377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,256,0.0026848000784715016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,256,0.014565333724021912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,512,0.007567999760309856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,128,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,128,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,256,0.0075989335775375364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,64,0.0022869333624839783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,128,0.007106133302052816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,64,0.014306132992108664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1536,32,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,65536,0.029262934128443403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1536,32,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,65536,0.04669440189997355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,16384,0.010807466506958009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,16384,0.02682773272196452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,2048,65536,0.12568426926930745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,12288,0.010016000270843506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,12288,0.024973867336908977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,12288,0.028242133061091107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,10240,0.01093226671218872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,10240,0.023545600970586143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,16384,0.03502613306045532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,8192,0.00928000013033549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,8192,0.022443733612696328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,10240,0.024537599086761473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,7168,0.00902506709098816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,7168,0.02093440095583598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,8192,0.020615466435750327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,6144,0.007930666704972585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,6144,0.019438934326171876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,6144,0.017382399241129557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,5120,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,5120,0.020055466890335084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,7168,0.01916159987449646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,4096,0.006635733445485433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,4096,0.018811732530593872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,5120,0.015698132912317912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,3584,0.005876266459623972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,4096,0.013329066832860312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,3584,0.017667200167973837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,3072,0.005469866593678792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,3072,0.01702186663945516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,3072,0.012053333719571431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,2560,0.005062399804592133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,3584,0.012984533111254373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,2560,0.0164874662955602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,2048,0.004644266764322917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,2048,0.01721386710802714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,2048,0.010365866621335347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,1536,0.003804799914360046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,2560,0.01130133370558421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,1536,0.01611306667327881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,1024,0.0033919999996821085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,1024,0.014935466647148132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,1024,0.008362666765848795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,768,0.0029674666623274487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,1536,0.009221333265304565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,768,0.014877866705258688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,512,0.002656000107526779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,512,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,512,0.007547733187675476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,768,0.008011733492215473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,256,0.0025909334421157838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,256,0.014627200365066529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1536,65536,0.12563412984212238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,256,0.007095466554164887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,128,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,128,0.01442346672217051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,64,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,64,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,1024,32,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,1024,32,0.014402133226394654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,128,0.007389866809050243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,65536,0.024780799945195518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,65536,0.042387199401855466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,16384,0.012001066406567892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,16384,0.026945066452026368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,12288,0.009946667154630025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,12288,0.023885866006215416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,12288,0.02834026614824931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,10240,0.008872532844543457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,10240,0.023146667083104453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,16384,0.0355189323425293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,8192,0.008167466521263123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,8192,0.022873600323994957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,8192,0.02107200026512146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,7168,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,7168,0.021076265970865884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,10240,0.024412800868352257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,6144,0.006749866902828217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,6144,0.020581332842508952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,7168,0.019049600760142008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,5120,0.007076266904671986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,6144,0.01741866668065389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,5120,0.01960853338241577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,4096,0.0063733334342638654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,4096,0.01874133348464966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,4096,0.013693867127100625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,3584,0.006275199850400289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,5120,0.01576746702194214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,3584,0.018331732352574667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,1024,65536,0.12146666844685872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,3072,0.0055178667108217875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,3072,0.017933867375055947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,2560,0.005058133105436961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,2560,0.016545066237449647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,3072,0.011710932850837708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,2560,0.011192533373832702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,2048,0.004631466666857402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,3584,0.012811733285586038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,1536,0.003804799914360046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,2048,0.015714133779207863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,1536,0.01618026693662008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,1024,0.003390933324893316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,1024,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,1536,0.009215999643007915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,768,0.003019733230272929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,1024,0.008593066533406576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,768,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,512,0.002619733413060506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,2048,0.010388267040252686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,512,0.014512000481287637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,256,0.0025781333446502686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,256,0.014865066607793173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,768,0.007991466422875721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,128,0.0021717332303524016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,512,0.007707733412583668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,128,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,256,0.007146666447321574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,64,0.002148266633351644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,768,32,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,128,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,64,0.014123732844988504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,65536,0.01790720025698344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,768,32,0.014360533157984415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,65536,0.041238399346669515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,16384,0.009578667084376017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,16384,0.02510400017102559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,12288,0.008331733445326488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,12288,0.02324159940083822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,12288,0.028472532828648884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,10240,0.007111466427644093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,10240,0.022765866915384927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,16384,0.035382401943206784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,8192,0.0070816000302632645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,8192,0.022785067558288574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,10240,0.024758400519688924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,7168,0.0066997334361076355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,7168,0.02078933318456014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,8192,0.02102933327356974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,6144,0.006270933151245117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,6144,0.019722666343053183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,7168,0.019466666380564372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,5120,0.006715733309586842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,5120,0.01980479955673218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,6144,0.017450666427612303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,4096,0.006691200037797292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,4096,0.017480534315109254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,5120,0.015783466895421348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,3584,0.00602346658706665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,3584,0.017962666352589925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,4096,0.013733333349227906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,3584,0.012885333100954691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,3072,0.005490133166313171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,3072,0.016590933005015053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,768,65536,0.12080106735229493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,2560,0.005116799970467886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,2560,0.01737066706021627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,3072,0.01209173301855723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,2048,0.004309333364168803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,2048,0.01597546637058258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,2560,0.011186133821805317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,1536,0.003881600002447764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,1536,0.017115734020868936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,1024,0.0033290666838486993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,2048,0.010401067137718201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,1024,0.015027200182278952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,1536,0.009136000275611877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,768,0.0032490665713946023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,768,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,512,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,512,0.014320000012715658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,1024,0.008368000388145447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,256,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,768,0.008067200084527333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,512,0.007914666831493378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,256,0.014730667074521383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,128,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,128,0.014104533195495605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,64,0.0022495999932289124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,128,0.007091199855009715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,256,0.007264000177383423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,512,32,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,64,0.013909332950909934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,65536,0.014551466703414917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,512,32,0.013831466436386108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,16384,0.006358399987220764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,65536,0.03504853248596192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,16384,0.02285760045051575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,12288,0.006410666803518932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,12288,0.02290880084037781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,12288,0.02810666759808858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,10240,0.00633493314186732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,16384,0.035427200794219973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,10240,0.022757333517074586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,8192,0.005927466849486033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,8192,0.021819732586542764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,10240,0.024842667579650878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,8192,0.02104426622390747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,7168,0.005551999807357788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,7168,0.02072746753692627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,6144,0.005930666625499725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,6144,0.020526933670043945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,7168,0.01949013272921244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,5120,0.006381866832574208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,6144,0.017717333634694417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,5120,0.018901334206263224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,4096,0.005938133100668589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,4096,0.01888426740964254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,5120,0.016081066926320393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,3584,0.005500799914201101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,4096,0.013705600301424661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,3584,0.016729599237442015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,3072,0.005452799797058106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,3072,0.01738986571629842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,3584,0.012916266918182373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,2560,0.005162666738033295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,2560,0.018499199549357095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,3072,0.012103466192881267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,2048,0.004664533336957296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,512,65536,0.12150826454162597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,2048,0.01634666621685028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,1536,0.0038399999340375268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,1536,0.015452800194422403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,2560,0.011212799946467083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,1024,0.0033845332761605583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,2048,0.010435199737548828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,1024,0.016316800316174825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,768,0.002979200085004171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,1536,0.009507200121879578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,768,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,512,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,1024,0.008403199911117553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,512,0.014501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,768,0.00788800021012624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,256,0.0022282667458057405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,256,0.014818132917086283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,128,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,512,0.00761599987745285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,256,0.0075082664688428235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,64,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,128,0.014447999993960061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,256,32,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,128,0.007158400118350982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,64,0.013636266191800436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,65536,0.010059733192125957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,256,32,0.014391466975212097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,65536,0.03169386585553487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,16384,0.005457066496213277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,16384,0.023065600792566934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,12288,0.005530666808287303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,12288,0.02282879948616028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,16384,0.03512746493021647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,256,65536,0.12028373082478841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,10240,0.005605333546797434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,10240,0.02076906760533651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,10240,0.024711465835571288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,12288,0.028388265768686933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,8192,0.005504000186920166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,8192,0.021169066429138184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,7168,0.005534933507442474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,7168,0.02053440014521281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,8192,0.020669867595036827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,6144,0.0057770664493242896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,6144,0.02028586665789286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,5120,0.0050346667567888895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,5120,0.019396267334620156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,7168,0.019008000691731773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,4096,0.005448533097902933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,4096,0.018756266434987387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,6144,0.017705599466959633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,3584,0.0052042668064435325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,5120,0.01569386621316274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,3584,0.018579200903574625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,3072,0.004771199822425842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,4096,0.01334933340549469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,3072,0.01760853330294291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,2560,0.0051242664456367494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,3584,0.012778666615486146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,2560,0.016498133540153503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,3072,0.01209493378798167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,2048,0.004833066463470459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,2048,0.01548693378766378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,1536,0.0038453333079814913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,2560,0.011183999975522359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,1536,0.015740799903869628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,2048,0.010431999961535137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,1024,0.003065599997838338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,1536,0.00918506681919098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,1024,0.01477226714293162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,768,0.0027776000400384264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,1024,0.008353066444396973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,768,0.015703466534614564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,65536,0.11984426975250244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,512,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,256,0.0022602667411168414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,512,0.014484266440073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,768,0.007915733257929485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,128,0.002145066608985265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,256,0.014750933647155762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,512,0.007565866907437642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,256,0.007524266839027405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,64,0.002242133269707362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,128,0.014724266529083253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,128,32,0.0022229333718617756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,128,128,128,0.006685866912206013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,64,0.013621333241462707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,65536,0.00751146674156189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,128,32,0.013462400436401368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,16384,0.005898666878541311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,12288,0.005482666691144307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,65536,0.030964267253875733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,16384,0.024294400215148927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,10240,0.005470933516820272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,12288,0.022756266593933105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,8192,0.005835733314355215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,10240,0.02072746753692627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,7168,0.005478399991989136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,6144,0.005063466727733612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,8192,0.021075199047724404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,7168,0.021705599625905354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,5120,0.004709333181381226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,6144,0.020244266589482626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,4096,0.004716800153255462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,3584,0.005061333378156027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,5120,0.019859200716018675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,4096,0.01898026665051778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,3072,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,3584,0.01757226586341858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,2560,0.005060266455014547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,2048,0.005047466854254405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,3072,0.017118932803471883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,2560,0.01680533289909363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,1536,0.003774933268626531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,2048,0.016306133071581522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,1024,0.0030741333961486817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,768,0.002625066787004471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,1024,0.015226667126019796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,1536,0.016226133704185484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,512,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,768,0.014840533336003622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,256,0.0021930667261282604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,128,0.002162133405605952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,512,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,256,0.014541866381963095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,64,0.0022101332743962605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,64,32,0.0021664001047611236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,128,0.014547200004259745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,64,0.015141333142916361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,65536,0.008728532989819845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,64,32,0.013461333513259888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,16384,0.005885866781075796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,12288,0.0054400001962979635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,65536,0.031990400950113934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,16384,0.024422399202982583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,10240,0.005513600011666616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,12288,0.022773333390553794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,8192,0.005487999816735586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,10240,0.0206112007300059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,7168,0.005397333204746247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,8192,0.020687999327977498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,6144,0.005052799979845682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,7168,0.01999680002530416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,5120,0.004693333307902018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,4096,0.0046304002404212955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,6144,0.020376533269882202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,3584,0.005043200155099233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,5120,0.018821332852045694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,4096,0.019425066312154134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,3072,0.004622933268547058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,2560,0.004692266881465912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,3584,0.01746986707051595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,3072,0.01663146714369456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,2048,0.004737066725889841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,1536,0.0038730666041374207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,2560,0.017174400885899863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,2048,0.016042666633923848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,1024,0.0030250666042168934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,768,0.0026709333062171934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,1536,0.014934399724006652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,1024,0.015447466572125753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,512,0.0026176000634829206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,256,0.00223786657055219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,768,0.015188266833623251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,512,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,128,0.0021802666286627453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,64,0.0018485333770513535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,256,0.015007999539375306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,128,0.014602667093276978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,128,32,32,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,64,0.014325333635012307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,128,32,32,0.013531733552614847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,16384,0.3854570706685384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,16384,0.2109674612681071
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,12288,0.16166399319966634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,12288,0.2769045193990072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,10240,0.21323199272155763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,10240,0.12447893619537354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,8192,0.1708192030588786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,8192,0.10649813016255696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,8192,0.06114346583684286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,7168,0.14501120249430338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,7168,0.0930506706237793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,10240,0.072325332959493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,6144,0.13091946442921956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,6144,0.08264426390329996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,12288,0.08503680229187012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,5120,0.10694613456726074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,5120,0.07189546426137289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,6144,0.04753706852595012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,4096,0.08877120018005372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,4096,0.06516480048497518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,7168,0.053496531645456945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,3584,0.07673172950744629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,3584,0.055572267373402914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,5120,0.041605333487192794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,3072,0.06682986418406169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,3072,0.04994986852010091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,16384,0.11304746468861897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,2560,0.056683735052744544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,2560,0.04530880053838094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,3584,0.031117866436640423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,2048,0.04733546574910481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,2048,0.040651734670003256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,4096,0.0342741330464681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,1536,0.036345601081848145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,1536,0.03279680013656616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,3072,0.02886613408724467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,1024,0.025784534215927125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,1024,0.025650133689244587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,2560,0.02563199996948242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,768,0.020081067085266115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,768,0.025524266560872394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,2048,0.02152000069618225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,512,0.013716266552607218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,512,0.02067306637763977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,1536,0.018278400103251137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,256,0.008357333143552144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,256,0.018041600783665977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,1024,0.014570666352907815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,128,0.005930666625499725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,128,0.016540799538294473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,768,0.012577066818873087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,64,0.004678399860858917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,64,0.018617600202560425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,65536,32,0.00673280010620753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,65536,32,0.01858560045560201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,128,0.008897067109743754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,65536,0.331608517964681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,65536,0.23309866587320963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,16384,0.09249813556671142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,16384,0.06531946659088135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,256,0.009256533781687419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,12288,0.06777706940968832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,12288,0.052439467112223304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,65536,512,0.010251733660697936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,10240,0.05789653460184733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,10240,0.04713279803593953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,10240,0.028806400299072266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,8192,0.04786133368810018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,8192,0.040453334649403885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,12288,0.033053867022196454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,7168,0.04046080112457275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,7168,0.03646399974822998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,16384,0.04158080021540324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,6144,0.036195198694864913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,6144,0.03503893216451009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,8192,0.024257065852483113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,5120,0.030462932586669923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,5120,0.031121067206064862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,7168,0.02160853346188863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,4096,0.02518293261528015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,4096,0.029234133164087933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,6144,0.019367466370264687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,3584,0.02218773365020752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,3584,0.026717867453893023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,4096,0.015025066335995993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,3072,0.019576533635457357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,3072,0.02393066684405009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,5120,0.017620267470677693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,2560,0.016764799753824867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,2560,0.02355626622835795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,2560,0.012082133690516155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,2048,0.014123732844988504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,2048,0.02151893377304077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,3072,0.013244799772898355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,1536,0.01083626647790273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,1536,0.019655466079711914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,3584,0.014070399602254233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,1024,0.008322133123874665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,1024,0.016583466529846193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,1024,0.00872320036093394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,768,0.0071274667978286745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,768,0.0164000004529953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,1536,0.00955839951833089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,512,0.0050346667567888895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,512,0.016571733355522155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,512,0.007550933460394542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,256,0.0035264000296592714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,256,0.01539306640625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,768,0.007947733501593272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,128,0.003068800022204717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,2048,0.011269332965215047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,128,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,64,0.002959999938805898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,64,0.014972800016403198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,16384,32,0.0030602666238943735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,16384,32,0.01460906664530436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,65536,0.2615423997243246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,65536,0.16558186213175458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,128,0.007133866846561432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,256,0.007160533467928569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,16384,0.07367253303527832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,16384,0.0577023983001709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,12288,0.05395306746164957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,12288,0.046528001626332596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,12288,0.030949334303538006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,10240,0.045672531922658285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,10240,0.041647998491923015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,16384,0.0390720009803772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,8192,0.03835839827855428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,8192,0.03640960057576497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,10240,0.027177600065867107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,7168,0.03254826664924622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,7168,0.03147413333257039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,8192,0.022372267643610635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,6144,0.029678932825724286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,6144,0.030395734310150146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,7168,0.02021226684252421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,5120,0.024401066700617473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,5120,0.027756800254185993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,6144,0.018665599822998046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,4096,0.02058773239453634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,4096,0.027142399549484254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,5120,0.01695893406867981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,3584,0.017642666896184288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,3584,0.0233130673567454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,4096,0.014939733346303306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,3072,0.01566933294137319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,3072,0.02370986739794413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,3584,0.013664000233014426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,2560,0.013598933815956116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,2560,0.021547732750574748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,3072,0.012910933295885722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,2048,0.011771733562151592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,2048,0.021078399817148843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,2560,0.011481600006421407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,1536,0.009252267082532246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,1536,0.018542933464050292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,1536,0.009611733754475911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,1024,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,1024,0.016537599762280784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,2048,0.010450133681297302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,768,0.005896533528963724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,768,0.015701333681742348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,768,0.008077866832415263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,512,0.004317866762479147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,512,0.014937600493431092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,1024,0.008724266290664673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,256,0.0033973333736260734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,256,0.015128533045450846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,256,0.007523199915885926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,128,0.002995199958483378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,128,0.014812800288200378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,128,0.007136000196139018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,64,0.0026261332134405774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,64,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,12288,32,0.002997333308060964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,12288,32,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,512,0.007542400062084198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,65536,0.2130357265472412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,65536,0.15403413772583008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,16384,0.06088106632232666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,16384,0.05152213176091512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,16384,0.037918933232625324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,12288,0.04428693453470866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,12288,0.04250986576080322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,12288,0.030105600754419964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,10240,0.03831679821014404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,10240,0.03844266732533773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,16384,65536,0.14817919731140136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,8192,0.03177493413289388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,8192,0.03294080098470052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,10240,0.025697066386540728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,7168,0.027287467320760088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,7168,0.030184533198674517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,8192,0.021951999266942343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,6144,0.02446933388710022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,6144,0.028486400842666626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,7168,0.02025173306465149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,5120,0.021041067441304524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,5120,0.025972266991933186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,12288,65536,0.13319146633148193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,4096,0.016982400417327882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,4096,0.024174932638804117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,5120,0.016953599452972413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,3584,0.014616533120473226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,3584,0.02250880002975464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,4096,0.014512000481287637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,3072,0.013733333349227906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,3072,0.022654932737350465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,6144,0.018641066551208497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,2560,0.011809066931406657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,2560,0.020708266894022623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,2560,0.011241599917411804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,2048,0.010372266173362732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,2048,0.01993280053138733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,3072,0.012220799922943115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,1536,0.008501332998275758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,1536,0.01761386593182882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,3584,0.013689600427945457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,1024,0.005881600081920624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,1024,0.01697280009587606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,1024,0.008447999755541485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,768,0.005162666738033295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,768,0.01569066643714905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,1536,0.009574400385220845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,512,0.004337066908677419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,512,0.016687999169031777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,2048,0.010422399640083313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,256,0.0034143999218940735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,256,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,768,0.007993599772453308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,128,0.0030933332939942675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,128,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,256,0.0075552001595497135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,64,0.003028266628583272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,64,0.015177599589029946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,10240,32,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,10240,32,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,128,0.007110400001207988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,65536,0.17071146965026857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,512,0.007658666869004567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,65536,0.1280191977818807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,16384,0.051191465059916175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,16384,0.045135998725891115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,12288,0.03615466753641765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,12288,0.0374944011370341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,12288,0.029165865977605183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,10240,0.03132906754811605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,10240,0.032959999640782674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,16384,0.037123199303944907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,8192,0.026420267422993977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,8192,0.029371732473373414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,10240,0.025868799289067584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,7168,0.02259626587231954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,7168,0.02746986746788025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,8192,0.02190399964650472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,6144,0.02029333313306173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,6144,0.025334399938583375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,7168,0.020142932732899986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,5120,0.01699840029080709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,5120,0.023124267657597862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,10240,65536,0.12879573504130046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,4096,0.014075733224550881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,4096,0.021870932976404824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,6144,0.018564265966415406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,3584,0.012520533800125123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,3584,0.020794665813446044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,4096,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,3072,0.010937600334485372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,3072,0.020592000087102255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,5120,0.016593066851298015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,2560,0.009659733374913533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,2560,0.02027946710586548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,3584,0.013302399714787801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,2048,0.008306133250395458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,2048,0.01797119975090027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,2560,0.011217066645622253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,1536,0.0070783997575442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,1536,0.01651946703592936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,2048,0.010075733065605164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,1024,0.005100800096988678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,1024,0.01700693368911743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,3072,0.0120469331741333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,768,0.003851733356714249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,768,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,768,0.007897600034872691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,512,0.0034495999415715536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,512,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,1536,0.009181867043177288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,1024,0.008711466193199157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,256,0.002919466545184453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,256,0.014418133099873862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,128,0.0026549334327379864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,128,0.01487573285897573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,128,0.0070602665344874065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,64,0.002616533388694127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,64,0.014446933070818582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,8192,32,0.002674133330583572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,8192,32,0.014541866381963095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,256,0.00751146674156189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,65536,0.146942933400472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,65536,0.11459519863128662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,16384,0.047464533646901445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,16384,0.04331946770350138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,512,0.00758186678091685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,12288,0.032295467456181844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,12288,0.03514026800791423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,12288,0.029475200176239013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,10240,0.027806933720906573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,10240,0.03150933384895325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,16384,0.03662613232930501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,8192,0.022755199670791627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,8192,0.028832000494003297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,10240,0.02555519938468933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,7168,0.021516799926757812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,7168,0.025663999716440837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,8192,0.02190933426221212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,6144,0.017832533518473307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,6144,0.024705066283543905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,7168,0.020011732975641884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,5120,0.01604693333307902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,5120,0.022668800751368203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,6144,0.01825066606203715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,4096,0.013351466258366904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,4096,0.021310933430989585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,5120,0.016343466440836587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,3584,0.011495467027028401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,3584,0.02103360096613566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,4096,0.014122666915257773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,3072,0.01032319962978363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,3072,0.019093332688013713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,3584,0.012826666235923767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,2560,0.009152000149091084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,2560,0.019372800985972084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,3072,0.012248532970746358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,2048,0.007938133180141449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,2048,0.01699840029080709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,2560,0.010790399710337321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,1536,0.006662400066852569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,1536,0.017563732465108235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,1536,0.00951039989789327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,1024,0.00470719983180364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,1024,0.017911465962727864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,2048,0.010168533523877461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,768,0.004277333120505015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,768,0.014829867084821067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,1024,0.008338133494059246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,512,0.00373333344856898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,768,0.007937066753705342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,512,0.014533332983652749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,256,0.00344106654326121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,256,0.014427733421325684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,256,0.007524266839027405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,128,0.0029824001093705496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,128,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,128,0.0071168000499407455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,64,0.0026389333109060925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,512,0.007890133559703827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,7168,32,0.002643200010061264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,64,0.014604799946149192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,7168,32,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,65536,0.12764693101247152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,65536,0.09890986283620198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,16384,0.040109864870707196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,16384,0.04272533257802327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,8192,65536,0.12678826649983724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,12288,0.031082665920257567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,12288,0.0319541335105896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,16384,0.037089065710703535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,10240,0.02733866572380066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,10240,0.028967465957005816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,12288,0.02885013421376546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,8192,0.021574399868647256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,8192,0.026664533217748004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,10240,0.02558186650276184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,7168,0.01935360034306844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,7168,0.02518506646156311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,8192,0.0218666672706604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,6144,0.016772266228993735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,6144,0.024589866399765015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,7168,0.019875200589497884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,5120,0.014040533701578775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,5120,0.02260479927062988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,6144,0.018641066551208497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,4096,0.011714133620262145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,4096,0.020835200945536293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,4096,0.013691733280817667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,3584,0.010709333419799804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,3584,0.01965120037396749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,5120,0.016360533237457276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,3072,0.009705600142478944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,3072,0.019233065843582153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,3584,0.013267200191815695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,2560,0.008428800106048583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,2560,0.01848640044530233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,3072,0.012115200360616047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,2048,0.007437866429487865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,2048,0.01686826745669047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,2048,0.010057600339253743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,1536,0.006153599917888641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,2560,0.01118293305238088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,1536,0.016764799753824867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,1024,0.004308266441027323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,1024,0.017058134078979492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,7168,65536,0.12619199752807617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,768,0.004011733333269755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,768,0.015780267119407655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,768,0.0079434668024381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,512,0.003435733417669932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,512,0.014844800035158793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,1024,0.008380800485610962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,256,0.002995199958483378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,256,0.014638933539390563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,256,0.007542400062084198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,512,0.007889066636562348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,128,0.0026186667382717133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,1536,0.009155199925104777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,128,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,64,0.002583466718594233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,6144,32,0.0025589334468046824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,64,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,6144,32,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,65536,0.11622933546702068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,128,0.007157333195209503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,65536,0.0928885300954183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,16384,0.03549013137817383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,16384,0.04100586573282878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,12288,0.027764266729354857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,12288,0.031100799640019734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,12288,0.029021867116292316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,10240,0.02527466615041097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,10240,0.02687679926554362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,16384,0.03588800032933553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,8192,0.01755946675936381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,8192,0.02480640014012655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,10240,0.025600000222524004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,7168,0.01525759994983673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,7168,0.024884267648061117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,8192,0.021851734320322672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,6144,0.013446399569511413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,6144,0.0232149342695872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,7168,0.019550933440526327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,5120,0.012016000350316365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,5120,0.02246933380762736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,6144,0.017753599087397258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,4096,0.01004266639550527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,4096,0.020878932873407998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,5120,0.01569386621316274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,3584,0.010177066922187806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,3584,0.01987733244895935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,4096,0.013833600282669067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,3072,0.009325866897900898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,3072,0.01872426668802897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,3584,0.012840533256530761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,2560,0.008261333405971526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,2560,0.019144533077875774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,3072,0.012103466192881267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,2048,0.007152000069618225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,2048,0.017015467087427773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,2560,0.011193600296974183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,1536,0.005970133344332377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,1536,0.01752106746037801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,2048,0.010066133737564088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,1024,0.004761599997679392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,1024,0.016427733500798545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,1536,0.009199999769528707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,768,0.0039061332742373147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,768,0.016451199849446617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,6144,65536,0.12618666489919025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,512,0.0035957333942254386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,1024,0.008403199911117553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,512,0.015638400117556253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,256,0.0031093334158261614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,768,0.008291199803352356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,256,0.014862933754920959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,128,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,128,0.014588800072669984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,256,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,128,0.007197866837183635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,64,0.0022079999248186748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,5120,32,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,64,0.014507733782132468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,5120,32,0.014284800489743552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,65536,0.09639360109965006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,65536,0.08212587038675943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,16384,0.028097067276636762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,512,0.007627733548482259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,16384,0.03532586495081584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,12288,0.023830399910608927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,12288,0.03088639974594116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,12288,0.02887679934501648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,10240,0.02068159977595011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,10240,0.02584106723467509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,16384,0.03589333295822143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,8192,0.015099733074506124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,8192,0.0237119992574056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,10240,0.025235199928283693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,7168,0.01325866679350535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,7168,0.02259413401285807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,8192,0.021285333236058555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,6144,0.012000000476837159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,6144,0.021389865875244142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,7168,0.019755733013153077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,5120,0.010413866241772969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,5120,0.01904319922129313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,6144,0.017461333672205606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,4096,0.008874666690826417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,4096,0.018651733795801796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,5120,0.015758933623631795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,3584,0.007929599781831106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,3584,0.01727893352508545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,4096,0.013869866728782654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,3072,0.007389866809050243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,3072,0.016936532656351724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,3584,0.012837333480517068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,2560,0.006277333199977875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,2560,0.016581333676973977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,2560,0.010805333654085796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,2048,0.00506986677646637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,3072,0.012156800429026286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,2048,0.01752426624298096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,1536,0.004320000112056732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,1536,0.016502400239308677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,1536,0.0092031995455424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,1024,0.0034645333886146545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,1024,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,2048,0.010316800077756245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,768,0.0032287999987602235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,768,0.01765120029449463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,1024,0.008355200290679932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,512,0.002996266633272171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,512,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,768,0.008008533219496409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,256,0.002677333354949951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,256,0.014562132954597472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,512,0.007550933460394542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,128,0.0025248001019159954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,128,0.01432213286558787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,128,0.007077333331108093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,64,0.0022442666192849475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,256,0.007176533341407776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,64,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,4096,32,0.0026047999660174055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,4096,32,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,65536,0.07882239818572997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,65536,0.07340266704559326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,16384,0.02595413327217102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,16384,0.03260053396224975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,5120,65536,0.12600213686625164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,12288,0.020438400904337566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,12288,0.02924586733182271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,12288,0.028899200757344562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,10240,0.017485866943995156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,10240,0.025484800338745117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,16384,0.035869868596394856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,8192,0.014910933375358582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,8192,0.02381333311398824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,8192,0.02123946746190389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,7168,0.014040533701578775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,7168,0.02289066712061564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,10240,0.02518293261528015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,6144,0.011564800143241882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,6144,0.020889600118001304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,7168,0.01948053240776062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,5120,0.009946667154630025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,5120,0.0200437327226003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,6144,0.017629865805308023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,4096,0.008918399612108868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,4096,0.018998400370279948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,4096,0.013920000195503235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,3584,0.008080000181992848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,3584,0.018711467583974205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,5120,0.01604373355706533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,3072,0.006846933563550313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,3072,0.017219199736913045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,3072,0.012038399775822956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,2560,0.005916800101598104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,2560,0.01765759984652201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,3584,0.013100799918174744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,2048,0.005212800204753875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,2048,0.01885546644528707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,4096,65536,0.1260693311691284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,1536,0.00443200021982193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,2560,0.011191466450691223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,1536,0.016717867056528727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,1024,0.00359253336985906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,1024,0.01497706671555837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,2048,0.01032960017522176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,768,0.0034613333642482757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,768,0.016863999764124553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,768,0.008382933338483174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,512,0.0030261332790056865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,1024,0.00846506655216217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,512,0.01477226714293162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,256,0.0027029333015282948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,256,0.014683733383814493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,1536,0.009130666653315227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,128,0.002584533393383026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,128,0.014200533429781595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,256,0.007558399935563405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,128,0.0070720002055168155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,64,0.0023029332359631857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3584,32,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,64,0.014442666371663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,512,0.007514666517575581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3584,32,0.014582399527231851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,65536,0.07015253702799479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,16384,0.02269866665204366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,65536,0.06888853708902995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,16384,0.030905600388844805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,12288,0.01780479947725932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,12288,0.027323732773462932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,12288,0.028742400805155437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,10240,0.015586133797963461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,10240,0.024961066246032716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,16384,0.035936001936594644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,8192,0.01360426644484202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,8192,0.023266132672627768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,10240,0.025252266724904375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,7168,0.01183786690235138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,7168,0.02145493427912394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,8192,0.021203200022379555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,6144,0.011016533772150675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,6144,0.021178666750590006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,7168,0.019061332941055296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,5120,0.009252267082532246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,5120,0.01946559945742289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,6144,0.017560533682505288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,4096,0.008040533463160197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,4096,0.018871466318766274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,5120,0.015683199961980185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,3584,0.007130666573842366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,3584,0.017780266205469766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,4096,0.013790933291117349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,3072,0.00678719977537791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,3584,0.012812800208727517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,3072,0.016708266735076905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,2560,0.006017066538333893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,2560,0.01686506668726603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3584,65536,0.12563412984212238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,2048,0.004826666911443075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,2560,0.01130133370558421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,2048,0.01687893271446228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,1536,0.0042912001411120095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,1536,0.015425067146619162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,3072,0.011662933230400085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,1024,0.003435733417669932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,1024,0.016314666469891867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,1024,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,768,0.003036800026893616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,768,0.014886400103569031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,1536,0.00920960009098053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,512,0.0028405333558718365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,512,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,768,0.007937066753705342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,256,0.0025802666942278544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,2048,0.010089600086212158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,256,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,128,0.002203733225663503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,512,0.007595733304818471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,128,0.013760000467300415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,256,0.007221333185831706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,128,0.006762666503588359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,64,0.002420266717672348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,64,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,3072,32,0.002348800003528595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,3072,32,0.01409386694431305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,65536,0.059246933460235594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,65536,0.06499413251876832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,16384,0.0206496000289917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,16384,0.02913706700007121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,12288,0.017375999689102174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,12288,0.026026666164398193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,12288,0.028534400463104247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,10240,0.014589866995811463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,10240,0.024913066625595094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,16384,0.0358517328898112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,8192,0.01249066690603892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,8192,0.02291839917500814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,10240,0.024875734249750772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,7168,0.01160533328851064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,7168,0.021549866596857706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,8192,0.020941867431004842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,6144,0.010214400291442872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,6144,0.020117332537968956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,6144,0.017364267508188883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,5120,0.008945066730181377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,5120,0.01872640053431193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,7168,0.01924053430557251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,4096,0.008024533092975617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,4096,0.019128533204396565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,5120,0.01570026675860087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,3584,0.007096533477306366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,4096,0.013790933291117349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,3584,0.0168938676516215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,3072,0.006265600025653839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,3072,0.01663040022055308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,3072,0.01162559986114502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,2560,0.0052704001466433205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,2560,0.01681813398996989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,3584,0.0126720001300176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,2048,0.004727466901143392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,2048,0.0166293332974116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,2560,0.010800000031789143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,1536,0.003824000060558319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,1536,0.015639467040697734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,2048,0.01014400025208791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,1024,0.003542399903138479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,1024,0.016286933422088624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,1024,0.008398933211962382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,768,0.003083733220895131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,1536,0.009196799993515015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,768,0.017254400253295898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,512,0.0026869334280490874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,512,0.014601600170135499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,768,0.007884799937407176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,512,0.007635200023651123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,256,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,256,0.014166399836540222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,128,0.0022698665658632917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,128,0.014225066701571146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,256,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,64,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,64,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2560,32,0.002198400100072225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,128,0.006932266553243001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2560,32,0.013757866621017457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,65536,0.04833279848098755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,65536,0.05740480025609335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,3072,65536,0.12583893140157063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,16384,0.0164874662955602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,16384,0.029165865977605183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,12288,0.014719999829928079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,12288,0.026881066958109538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,12288,0.02813226580619812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,10240,0.012937600413958231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,10240,0.02791680097579956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,16384,0.03567999998728434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,8192,0.011322666207949321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,8192,0.0231989324092865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,10240,0.024510933955510458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,7168,0.010648533701896667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,7168,0.023067732652028404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,8192,0.0210207998752594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,6144,0.009640533725420635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,6144,0.020951465765635172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,7168,0.019189333915710448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,5120,0.01202453374862671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,5120,0.021309866507848104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,6144,0.017339734236399333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,4096,0.009429333607355754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,4096,0.019740800062815346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,4096,0.013784533739089966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,3584,0.008397866288820903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,3584,0.017335466543833413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,5120,0.015801599621772765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,3072,0.007896533111731212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,3072,0.01709653337796529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,3584,0.012824533383051553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,2560,0.00673173318306605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,3072,0.012158933281898498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,2560,0.016908800601959227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,2048,0.006054399907588959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,2048,0.01581013302008311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,2560,0.011293866237004598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,1536,0.004334933559099833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,1536,0.01585813363393148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,2048,0.010461866855621338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,1024,0.0034175999462604523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,1024,0.01460693379243215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,1536,0.009283199906349182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,768,0.0030634666482607523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2560,65536,0.1255786657333374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,768,0.014665599664052328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,512,0.002902399996916453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,512,0.01453013320763906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,512,0.007924266656239827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,768,0.007972266773382823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,256,0.002570666621128718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,256,0.014510933558146158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,1024,0.008350933591524761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,128,0.002218666672706604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,128,0.01390720009803772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,64,0.002181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,64,0.014678399761517844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,256,0.00720000018676122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,2048,32,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,128,0.0071829333901405334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,65536,0.03998399972915649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,2048,32,0.014588800072669984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,16384,0.013906133174896241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,65536,0.051577599843343105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,16384,0.029208532969156903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,12288,0.011763200163841248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,12288,0.025177599986394246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,12288,0.028076799710591634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,10240,0.01083626647790273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,10240,0.02407039999961853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,16384,0.03514453172683716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,8192,0.010308266679445902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,8192,0.02262186606725057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,10240,0.02472320000330607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,7168,0.009807999928792317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,7168,0.020778665939966835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,8192,0.020667733748753865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,6144,0.008847999572753906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,6144,0.01941653291384379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,7168,0.019181867440541588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,5120,0.008328533172607422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,5120,0.019885865847269694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,6144,0.01745706597963969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,4096,0.00912000040213267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,4096,0.018062933286031087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,5120,0.01577600042025248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,3584,0.006320000191529592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,3584,0.017953066031138103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,4096,0.013658666610717773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,3072,0.005582933127880096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,3072,0.01884053349494934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,3584,0.012877866625785828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,2560,0.005102933446566264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,2560,0.016724266608556113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,3072,0.012072533369064331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,2048,0.004655999938646952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,2048,0.01541973352432251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,2560,0.01116373340288798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,1536,0.003927466770013174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,1536,0.016938666502634682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,2048,0.010384000341097514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,1024,0.0034901333351929987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,1024,0.014491732915242514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,1536,0.009107200304667155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,768,0.0030421334008375804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,768,0.01649173299471537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,1024,0.008730666836102803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,512,0.0026709333062171934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,512,0.014551466703414917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,512,0.007540266712506611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,768,0.008278400202592214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,256,0.002603733291228612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,2048,65536,0.12504746913909912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,256,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,128,0.002250666668017705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,128,0.014513066411018372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,64,0.002217599997917811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,64,0.014641066392262777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,128,0.007134933272997539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,256,0.007106133302052816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1536,32,0.0022410665949185687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,65536,0.028459733724594115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1536,32,0.013585066795349121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,65536,0.04738346735636394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,16384,0.012102400263150532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,16384,0.0290613333384196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,12288,0.01032426655292511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,12288,0.025076266129811602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,12288,0.028016000986099243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,10240,0.009693866968154908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,10240,0.02484906713167826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,16384,0.03505280017852783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,8192,0.011210667093594869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,8192,0.02291626731554667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,10240,0.024401066700617473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,7168,0.01039466659228007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,7168,0.02077546715736389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,7168,0.018963199853897095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,6144,0.009617066383361817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,6144,0.019822933276494346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,8192,0.020713599522908528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,5120,0.008769067128499348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,5120,0.018862932920455933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,5120,0.015715199708938598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,4096,0.007904000083605449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,4096,0.018644267320632936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,6144,0.01730133295059204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,3584,0.006226133306821187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,3584,0.01909546653429667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,3584,0.012937600413958231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,4096,0.01362559994061788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,3072,0.0055071999629338585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,3072,0.019027199347813925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,2560,0.005038933455944061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,2560,0.016745599110921223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,2560,0.010796800255775452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,2048,0.00544106662273407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,3072,0.01204266647497813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,2048,0.017443199952443443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,1536,0.004817066589991251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,1536,0.01628373364607493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,1536,0.009139200051625569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,1024,0.0038250667353471124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,1024,0.01634986698627472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,2048,0.00992746651172638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,768,0.00342399999499321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,768,0.017411200205485027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,768,0.008274133503437042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,1024,0.008522666494051616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,512,0.0030464000999927522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,512,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,256,0.0025621332228183747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,256,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,512,0.007863466441631318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,128,0.0026047999660174055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,256,0.007068799932797749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,128,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,64,0.002478933334350586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,128,0.0070826664566993715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,64,0.014632532993952433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,1024,32,0.0024458666642506914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,65536,0.024358399709065757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,1024,32,0.01467519998550415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,65536,0.03934079806009928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,16384,0.014906666676203408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,16384,0.02655893365542094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1536,65536,0.1243168036142985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,12288,0.012401066223780314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,12288,0.024661332368850708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,12288,0.0280074675877889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,10240,0.010829866925875346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,10240,0.023406932751337685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,16384,0.035206401348114015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,8192,0.009749333063761394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,8192,0.02135146657625834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,10240,0.024337067206700643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,7168,0.009144533673922222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,7168,0.020483199755350748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,8192,0.02064639925956726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,6144,0.008508800466855367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,6144,0.019016534090042114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,7168,0.019042134284973145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,5120,0.007923200229803721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,5120,0.018632533152898152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,6144,0.017407999436060587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,4096,0.00684799998998642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,4096,0.017029333114624023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,5120,0.015683199961980185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,3584,0.005876266459623972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,3584,0.017043199141820273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,1024,65536,0.1212661345799764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,3072,0.0058005332946777345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,4096,0.013378133376439413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,3072,0.018101332585016887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,2560,0.005268266797065735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,2560,0.017117865880330405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,3584,0.012870400150616964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,2048,0.004285866518815359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,2048,0.01665493349234263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,3072,0.01209386686484019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,1536,0.0038794666528701783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,2560,0.010921600461006164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,1536,0.016871466239293417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,1024,0.0033728001018365227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,1024,0.015261866648991904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,2048,0.009962667028109233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,768,0.0029781334102153777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,768,0.016182399789492288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,1536,0.009266133109728496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,1024,0.008788266777992248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,512,0.0025962665677070618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,512,0.014062933127085366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,768,0.00754559983809789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,256,0.0025962665677070618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,256,0.014252799749374389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,512,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,128,0.0022912000616391497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,256,0.007067733506361644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,128,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,64,0.002299733211596807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,128,0.006683733562628429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,768,32,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,64,0.013916800419489542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,65536,0.01731520096460978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,768,32,0.013809067010879517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,65536,0.03766186634699504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,16384,0.011635200182596842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,16384,0.024895999828974405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,12288,0.009331199526786804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,12288,0.023116799195607503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,16384,0.03505386511484782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,10240,0.008452266454696655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,10240,0.022530132532119752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,12288,0.028090665737787884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,8192,0.007539199789365132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,8192,0.02103360096613566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,10240,0.02444266676902771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,7168,0.00716480016708374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,7168,0.0205567995707194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,8192,0.020662399133046468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,6144,0.006322133541107178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,6144,0.019646932681401573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,7168,0.01946559945742289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,5120,0.006763733426729838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,6144,0.01739413340886434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,5120,0.019230933984120686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,4096,0.006681600213050842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,4096,0.018528000513712565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,5120,0.015757866700490317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,4096,0.013723733027776084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,3584,0.006175999840100607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,3584,0.017734400431315103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,3072,0.00550186683734258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,3072,0.0168778657913208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,768,65536,0.1206997315088908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,2560,0.005115733544031779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,2560,0.01771519978841146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,3584,0.012846933801968894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,2048,0.004277333120505015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,3072,0.01202880044778188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,2048,0.01771519978841146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,1536,0.0038560000558694207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,1536,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,2560,0.01107413371404012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,1024,0.003366400053103765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,2048,0.010377599795659383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,1024,0.015158399939537048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,1536,0.009201066692670186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,768,0.0030432000756263735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,768,0.01488746702671051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,512,0.002598399917284648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,512,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,1024,0.00832533339659373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,768,0.007916800181070964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,256,0.002476799984773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,256,0.014353066682815552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,128,0.002243199944496155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,512,0.007495466868082683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,256,0.0071722666422526045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,128,0.014439466595649719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,64,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,512,32,0.0021898667017618816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,128,0.006746666630109151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,64,0.014617600043614707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,65536,0.014660267035166421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,512,32,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,16384,0.007166933516661327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,65536,0.035079467296600345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,16384,0.022829866409301756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,12288,0.006693333387374878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,12288,0.02246933380762736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,12288,0.02808213432629903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,10240,0.006286933521429698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,10240,0.022499199708302817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,16384,0.03481066624323527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,8192,0.005996799965699514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,8192,0.022196267048517862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,10240,0.02480319937070211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,7168,0.005491200089454651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,8192,0.020564266045888267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,7168,0.020642133553822835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,6144,0.006301866471767425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,6144,0.020003199577331543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,7168,0.019011199474334717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,5120,0.006683733562628429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,6144,0.01772159934043884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,5120,0.019760000705718993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,4096,0.006292266647020976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,4096,0.018794665733973183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,5120,0.015698132912317912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,3584,0.0058783998092015585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,4096,0.013706666231155396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,3584,0.018447999159495035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,512,65536,0.12025600274403889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,3072,0.00556160012880961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,3072,0.017079466581344606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,3584,0.012935466567675271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,2560,0.005576533575852713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,2560,0.017103999853134155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,3072,0.012056533495585125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,2048,0.005064533154169718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,2048,0.016199466586112977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,2560,0.011307733257611592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,1536,0.00384853333234787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,1536,0.016039466857910155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,2048,0.010356266299883525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,1024,0.003049599876006444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,1024,0.014661332964897156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,1536,0.009628799557685853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,768,0.003066666672627131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,1024,0.008379733562469483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,768,0.014636799693107605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,512,0.002872533351182938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,768,0.0074976002176602675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,512,0.01470186710357666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,256,0.0022762666145960493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,512,0.007501866420110066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,256,0.014573867122332254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,256,0.00714026689529419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,128,0.0022677332162857056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,64,0.0022495999932289124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,128,0.014327466487884521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,128,0.006698666512966156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,64,0.014392532904942832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,256,32,0.0022730665902296705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,65536,0.009955199559529622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,256,32,0.013497599959373474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,16384,0.005505066613356272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,65536,0.03134613235791524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,16384,0.022654932737350465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,12288,0.005482666691144307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,12288,0.02146453261375427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,256,65536,0.12047786712646484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,10240,0.00573333352804184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,12288,0.027970133225123088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,10240,0.020278400182723998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,16384,0.03498880068461101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,8192,0.005579733351866404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,8192,0.020923733711242676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,7168,0.005550933380921682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,7168,0.020807466904322305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,10240,0.024280534187952677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,6144,0.005053866902987162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,8192,0.020670932531356812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,6144,0.01904746691385905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,5120,0.005046399931112925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,7168,0.019081600507100425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,5120,0.020524799823760986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,4096,0.005393066505591074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,4096,0.019074134031931558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,6144,0.017356799046198527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,3584,0.0050784001747767125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,5120,0.01579839984575907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,3584,0.017282134294509886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,3072,0.005064533154169718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,4096,0.01372160017490387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,3072,0.016158933440844216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,2560,0.005111466844876607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,3584,0.012814933061599731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,2560,0.01796906590461731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,2048,0.005054933329423269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,3072,0.012029866377512615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,2048,0.01634666621685028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,1536,0.0038346665600935614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,2560,0.010825600226720173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,2048,0.010024533669153849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,1536,0.016109866897265117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,1024,0.0029898665845394133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,1024,0.015347199638684592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,768,0.0030517332255840302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,1536,0.009215999643007915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,1024,0.008291199803352356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,768,0.01639253298441569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,512,0.002593066543340683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,768,0.007929599781831106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,512,0.013848533233006796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,256,0.0022837333381175995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,512,0.0075082664688428235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,256,0.013153066237767538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,128,0.0022111999491850535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,256,0.006734933455785115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,64,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,128,0.013130666812260947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,128,0.007043200234572093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,128,32,0.00210453321536382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,64,0.014595199624697366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,65536,0.009187199672063192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,128,32,0.012557866175969443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,16384,0.0058890665570894875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,12288,0.005457066496213277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,65536,0.031222399075826007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,10240,0.005859200159708659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,16384,0.022692267100016275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,12288,0.020755199591318767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,10240,0.020805333058039346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,8192,0.005515733361244201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,7168,0.005475200215975444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,8192,0.019870932896931967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,6144,0.005037866532802582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,96,128,65536,0.1199338674545288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,7168,0.01869973341623942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,5120,0.005163733164469401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,6144,0.018747733036677042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,4096,0.004755199948946635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,5120,0.019037866592407228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,3584,0.005195733408133189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,4096,0.01844373345375061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,3072,0.005108266572157542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,3584,0.017030400037765504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,2560,0.00513919989267985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,3072,0.01654293338457743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,2048,0.004747733473777771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,1536,0.0038165333370367683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,2560,0.016035200158754984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,2048,0.016849066813786825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,1024,0.003031466652949651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,768,0.0026933332284291584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,1536,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,1024,0.014434132973353067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,512,0.0026858667532602947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,768,0.014644267161687216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,256,0.0023007998863856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,512,0.013778133193651834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,128,0.0023039999107519788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,64,0.002229333420594533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,256,0.012940800189971924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,128,0.012784000237782797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,64,32,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,32,0.013275733590126038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,64,64,0.014121599992116294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,65536,0.00886079967021942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,16384,0.0063967997829119366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,65536,0.030896000067392987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,12288,0.0055071999629338585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,16384,0.02264639933904012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,10240,0.006042666733264923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,8192,0.005924266576766968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,12288,0.02116480072339376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,7168,0.0054058666030565895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,10240,0.020665599902470907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,6144,0.005138133466243744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,8192,0.019925334056218467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,7168,0.018921599785486857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,5120,0.005221333106358846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,4096,0.004704000055789947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,6144,0.01943999926249186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,5120,0.018962132930755615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,3584,0.004671999812126159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,3072,0.00462719996770223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,4096,0.01865066687266032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,3584,0.0167797327041626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,2560,0.005052799979845682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,2048,0.004699733356634776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,3072,0.01656000018119812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,2560,0.018334933121999106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,1536,0.003458133339881897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,1024,0.003018666555484136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,2048,0.016858667135238647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,1536,0.01597546637058258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,768,0.003011200080315272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,512,0.002609066665172577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,1024,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,768,0.015051733454068503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,256,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,128,0.002205866575241089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,512,0.014579199751218162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,256,0.01297706663608551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,64,0.002209066599607468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,96,32,32,0.0021685334543387097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,128,0.012850133577982583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,64,0.014316800236701965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,96,32,32,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,16384,0.20634026527404786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,12288,0.27468051910400393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,16384,0.36899305979410807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,12288,0.15835413932800294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,10240,0.2107423941294352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,10240,0.12448853651682537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,8192,0.17246613502502442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,8192,0.10394986470540364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,8192,0.06345066626866659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,7168,0.1470570723215739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,7168,0.09196480115254721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,10240,0.07522772947947184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,6144,0.12985599835713704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,6144,0.08114453156789145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,12288,0.08420586585998535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,5120,0.10715626875559489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,5120,0.07050346533457438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,6144,0.05094826618830363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,4096,0.08758613268534342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,4096,0.06258879899978638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,7168,0.057366398970286045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,3584,0.0766048034032186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,3584,0.05464533170064291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,5120,0.04257599910100301
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,3072,0.06616640090942383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,3072,0.04954986572265625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,16384,0.1117791970570882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,2560,0.05609386761983236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,2560,0.044494934876759845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,3584,0.031726932525634764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,2048,0.04761173327763875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,2048,0.03898986577987671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,4096,0.03700266679128011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,1536,0.0355189323425293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,1536,0.03265173236529033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,3072,0.028602667649586994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,1024,0.02533973256746928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,1024,0.025536000728607178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,2560,0.02482773264249166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,768,0.019703465700149535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,768,0.023277866840362548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,2048,0.021571199099222817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,512,0.013758933544158936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,512,0.02062293291091919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,1536,0.017987199624379478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,256,0.008424533406893413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,256,0.017643733819325765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,256,0.009256533781687419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,128,0.005564799904823304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,128,0.0165802667538325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,128,0.008802133798599242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,64,0.004628266890843709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,64,0.016663466890652977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,65536,32,0.005469866593678792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,65536,32,0.018663465976715088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,65536,0.3090165456136068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,65536,0.2135178724924723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,512,0.01055680016676585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,16384,0.08685973485310873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,16384,0.06131306489308676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,768,0.012574932972590127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,12288,0.062609068552653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,12288,0.052077865600585936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,65536,1024,0.014553599556287131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,10240,0.053342934449513754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,10240,0.04490240017573039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,10240,0.029021867116292316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,8192,0.04345493316650391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,8192,0.03879679838816325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,12288,0.03290560046831767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,7168,0.03769599994023641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,7168,0.03596906661987305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,16384,0.04194133281707764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,6144,0.03339200019836426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,6144,0.03228373328844707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,8192,0.024142932891845704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,5120,0.028463999430338543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,5120,0.030155734221140547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,7168,0.021523199478785195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,4096,0.023085866371790567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,4096,0.026609067122141523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,5120,0.017395200332005818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,3584,0.020707199970881142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,3584,0.023909332354863484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,6144,0.019160532951354982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,3072,0.018051199118296304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,3072,0.023176532983779908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,4096,0.014941866199175516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,2560,0.015479466319084168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,2560,0.021862399578094483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,3584,0.014150399963061014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,2048,0.01244160036245982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,2048,0.020665599902470907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,2560,0.012167466680208842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,1536,0.010056533416112264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,1536,0.01874986688296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,3072,0.013208533326784769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,1024,0.007590400179227193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,1024,0.015460266669591268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,1024,0.008388266960779826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,768,0.006387199958165486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,768,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,1536,0.009397332866986591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,512,0.0044159998496373495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,512,0.01648853321870168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,768,0.007892266909281413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,256,0.0034122665723164878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,256,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,512,0.007725866635640462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,128,0.0030794667700926462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,128,0.014667733510335287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,256,0.007146666447321574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,64,0.0026122666895389557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,64,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,16384,32,0.002625066787004471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,128,0.007147733370463054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,16384,32,0.014498133460680643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,65536,0.24826879501342775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,2048,0.01086293359597524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,65536,0.16537173589070636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,16384,0.07378666400909424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,16384,0.05701653162638346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,12288,0.05375466744105021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,12288,0.04518186648686727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,12288,0.030845866600672407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,10240,0.04575999975204468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,10240,0.04139413436253865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,16384,0.03915946483612061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,8192,0.03876373370488485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,8192,0.03575573364893596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,10240,0.02680319945017497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,7168,0.03217173417409261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,7168,0.033218133449554446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,8192,0.022360533475875854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,6144,0.029288534323374433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,6144,0.028924800952275592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,7168,0.02021226684252421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,5120,0.024578134218851723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,5120,0.026961066325505573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,6144,0.018662399053573607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,4096,0.02033920089403788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,4096,0.024833067258199056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,5120,0.016982400417327882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,3584,0.017783466974894205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,3584,0.025543467203776045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,4096,0.01493119994799296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,3072,0.015140266219774882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,3072,0.023101866245269775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,3584,0.013858133554458618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,2560,0.013293866316477457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,2560,0.02068586746851603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,3072,0.01290773351987203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,2048,0.011437867085138958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,2048,0.020257065693537392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,2560,0.011591466267903645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,1536,0.009191466371218364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,1536,0.01882773240407308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,2048,0.010537599523862202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,1024,0.007218133409818013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,1024,0.016821332772572837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,1536,0.00958293378353119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,768,0.0055071999629338585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,768,0.014813866217931113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,1024,0.008787199854850769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,512,0.004309333364168803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,512,0.015479466319084168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,768,0.008297599852085114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,256,0.0034143999218940735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,256,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,256,0.007100800176461537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,128,0.0030634666482607523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,128,0.014699733257293702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,128,0.007136000196139018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,64,0.0026496000587940215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,64,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,12288,32,0.003036800026893616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,12288,32,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,512,0.007625600198904674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,65536,0.211844269434611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,65536,0.15009066263834636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,16384,0.062386135260264076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,16384,0.049955201148986814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,16384,0.03795093297958374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,12288,0.04442773262659709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,12288,0.041341865062713624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,12288,0.02980159918467204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,10240,0.038303999106089275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,10240,0.037249068419138595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,16384,65536,0.14661760330200196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,8192,0.031795199712117514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,8192,0.031896533568700154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,10240,0.02600746750831604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,7168,0.02739306688308716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,7168,0.02986666758855184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,8192,0.02234026590983073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,6144,0.0244704008102417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,6144,0.02754240036010742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,7168,0.0202890674273173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,5120,0.020757333437601725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,5120,0.024784000714619954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,12288,65536,0.13300800323486328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,4096,0.017164800564448038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,4096,0.02579413255055745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,6144,0.01858453353246053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,3584,0.01611199975013733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,3584,0.021618133783340453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,5120,0.016567466656366985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,3072,0.013174399733543396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,3072,0.021673599878946938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,4096,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,2560,0.011660800377527872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,2560,0.021772799889246623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,3584,0.013310933113098144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,2048,0.01037013332049052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,2048,0.019486933946609497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,3072,0.012422399719556172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,1536,0.008056533336639405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,1536,0.017249067624409996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,2560,0.011252267162005107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,1024,0.006278400123119354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,1024,0.018147200345993042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,1536,0.009196799993515015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,768,0.00477866679430008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,768,0.0164490669965744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,1024,0.00837546686331431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,512,0.004192000130812327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,512,0.016203733285268147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,2048,0.010364799698193868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,256,0.0034229333202044168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,256,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,768,0.008004266520341237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,128,0.0029887999097506206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,256,0.007085866729418437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,128,0.015156267086664834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,64,0.0026421333352724713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,64,0.014524799585342408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,10240,32,0.002712533374627431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,10240,32,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,65536,0.17166825930277507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,512,0.00749120016892751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,128,0.007100800176461537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,65536,0.1254965305328369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,16384,0.047312001387278244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,16384,0.04327359994252523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,12288,0.035359998544057206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,12288,0.036433064937591554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,12288,0.029304534196853638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,10240,0.030534400542577105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,10240,0.03161493341128031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,16384,0.037050668398539224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,8192,0.025256532430648803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,8192,0.028727465867996217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,10240,0.025568000475565594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,7168,0.022030933698018392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,7168,0.02574186722437541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,8192,0.021846399704615275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,6144,0.019836799303690592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,6144,0.02485760052998861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,7168,0.02026346723238627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,5120,0.016495999693870545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,5120,0.023308799664179484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,10240,65536,0.12797973155975342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,4096,0.013714133699735006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,4096,0.02264960010846456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,6144,0.01829973260561625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,3584,0.012218667070070903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,3584,0.020730666319529214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,5120,0.016622933745384216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,3072,0.010898133118947346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,3072,0.019454934199651084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,3584,0.013313066959381104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,2560,0.009629866480827332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,2560,0.019106133778889974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,4096,0.014575999975204468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,2048,0.008329600095748901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,2048,0.017822933197021485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,3072,0.012493866682052612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,1536,0.007042133311430614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,1536,0.017773866653442383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,2560,0.011352533102035522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,1024,0.004748799900213877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,1024,0.015409066279729208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,2048,0.010003200173377991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,768,0.003894400099913279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,768,0.014526933431625366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,1536,0.009266133109728496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,512,0.0035445332527160645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,512,0.01458346645037333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,768,0.00795413355032603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,256,0.0029386666913827257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,256,0.01454080045223236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,1024,0.008438400427500407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,128,0.0026911998788515727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,128,0.014594133694966635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,512,0.007968000074227651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,128,0.0067445332805315655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,64,0.0024480000138282776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,8192,32,0.002632533262173335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,64,0.01442026694615682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,8192,32,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,256,0.007066666583220164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,65536,0.15507413546244303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,16384,0.045902931690216066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,65536,0.1107413371404012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,16384,0.04581760168075562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,12288,0.032763733466466265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,12288,0.03452906608581543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,12288,0.028858667612075804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,10240,0.02831573287645976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,10240,0.03156053423881531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,16384,0.03717546860376994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,8192,0.023601067066192628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,8192,0.026885332663853963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,10240,0.025546665986378985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,7168,0.0201365331808726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,7168,0.02601813276608785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,8192,0.021898667017618813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,6144,0.018193066120147705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,6144,0.02534293333689372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,7168,0.01997119983037313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,5120,0.015491200486818948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,5120,0.022848000129063926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,6144,0.01843520005544027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,4096,0.012869333227475485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,4096,0.02106026609738668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,5120,0.016387200355529784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,3584,0.011738666892051696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,3584,0.02147093415260315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,4096,0.01409280002117157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,3072,0.010412800312042236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,3072,0.02004479964574178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,3584,0.012849066654841104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,2560,0.009253333012262981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,2560,0.018627200524012247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,3072,0.012295466661453248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,2048,0.007928533355395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,2048,0.016525866587956746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,2560,0.011248000462849935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,1536,0.006469333171844482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,1536,0.01611199975013733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,2048,0.010328533252080281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,1024,0.004653866589069367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,1024,0.014738133549690247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,1536,0.009541333715120951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,768,0.0038762666285037995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,768,0.017915733655293784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,1024,0.008755200107892354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,512,0.0034133332471052804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,512,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,512,0.00788373351097107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,256,0.0029653333127498626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,256,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,768,0.007990399996439617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,128,0.0025557334224383037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,128,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,128,0.0067221333583196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,64,0.002584533393383026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,64,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,7168,32,0.002794666588306427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,256,0.007134933272997539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,7168,32,0.014386133352915446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,65536,0.12294507026672363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,65536,0.09388373692830404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,16384,0.0388864000638326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,16384,0.04132800102233887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,8192,65536,0.12666133244832356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,12288,0.029806933800379437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,12288,0.035946667194366455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,16384,0.036245334148406985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,10240,0.026708267132441205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,10240,0.02908160090446472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,12288,0.028864000240961713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,8192,0.02244159976641337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,8192,0.027873067061106364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,10240,0.02555626630783081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,7168,0.01919680039087931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,7168,0.02595733404159546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,8192,0.021814399957656862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,6144,0.017332265774408974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,6144,0.024807467063268026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,7168,0.019847466548283895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,5120,0.014846932888031007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,5120,0.022948267062505086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,6144,0.01781866749127706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,4096,0.012940800189971924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,4096,0.022910932699839272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,5120,0.016261333227157594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,3584,0.011554132898648579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,3584,0.019922133286794028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,4096,0.013723733027776084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,3072,0.01058666706085205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,3072,0.01911146640777588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,3584,0.012853333353996277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,2560,0.009595732887585957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,2560,0.017476266622543334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,3072,0.012114133437474568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,2048,0.007437866429487865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,2048,0.016680532693862916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,2560,0.011188266674677531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,1536,0.006617600222428639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,1536,0.016173866391181946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,2048,0.010041600465774536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,1024,0.004706133405367533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,1024,0.017112533251444496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,7168,65536,0.12626559734344484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,768,0.003905066599448522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,768,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,768,0.007899733384450276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,1536,0.009237333138783773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,512,0.0037237333754698435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,512,0.01467626690864563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,256,0.0030879999200503034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,256,0.014533332983652749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,1024,0.008844799796740214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,128,0.002661333233118057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,128,0.014739200472831726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,128,0.00717439999183019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,64,0.0026464000344276427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,256,0.0071829333901405334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,6144,32,0.00266239990790685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,64,0.014651733636856078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,6144,32,0.014532267053922018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,65536,0.1059765338897705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,512,0.007559466858704884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,65536,0.09101760387420654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,16384,0.03356159925460815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,16384,0.03827306826909383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,12288,0.026792534192403156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,12288,0.031294933954874676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,12288,0.02882133324940999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,10240,0.02318399945894877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,10240,0.028894933064778645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,16384,0.03586560090382894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,8192,0.02102400064468384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,8192,0.025221333901087446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,10240,0.025463465849558515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,7168,0.018233599265416463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,7168,0.022768000761667885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,8192,0.021601066986719767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,6144,0.013806933164596557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,6144,0.024207999308904014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,7168,0.019820799430211387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,5120,0.012055466572443645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,5120,0.02127893368403117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,6144,0.01792959968249003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,4096,0.010052266716957092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,4096,0.02158613403638204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,5120,0.016116266449292503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,3584,0.009194667140642803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,3584,0.019013333320617675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,4096,0.013640532890955607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,3072,0.010321066776911417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,3072,0.02145386735598246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,3584,0.012827733159065246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,2560,0.009101866682370504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,2560,0.016596266627311708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,3072,0.012446932991345723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,2048,0.007090133428573608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,2048,0.017194666465123496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,2560,0.011195733149846395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,1536,0.00631573349237442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,1536,0.016426666577657064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,6144,65536,0.1262933333714803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,1024,0.0047647997736930845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,1024,0.014776532848676046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,2048,0.010020266969998677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,768,0.004276266694068909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,1536,0.009142399827639262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,768,0.014918399850527444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,512,0.003491200009981791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,512,0.014781866470972696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,512,0.007493333518505096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,256,0.0030517332255840302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,256,0.014735999703407287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,1024,0.008762666583061218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,128,0.0025898667673269907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,128,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,768,0.007940266529719036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,64,0.0025909334421157838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,64,0.014521599809328715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,5120,32,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,128,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,5120,32,0.01442346672217051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,256,0.007158400118350982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,65536,0.09209493001302084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,65536,0.07529599666595459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,16384,0.026687999566396076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,16384,0.03368320067723592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,12288,0.02267520030339559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,12288,0.02993173400561015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,12288,0.02846933404604594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,10240,0.019826134045918785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,10240,0.025050665934880572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,16384,0.03600533405939738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,8192,0.014989866813023885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,8192,0.0236245334148407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,10240,0.025170133511225386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,7168,0.013201066851615905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,7168,0.021780266364415487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,8192,0.021109332640965782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,6144,0.011865599950154623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,6144,0.021769599119822184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,7168,0.01944533387819926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,5120,0.010429867108662923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,5120,0.019359999895095827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,6144,0.01740480065345764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,4096,0.008756267031033833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,4096,0.018539732694625853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,5120,0.016132266322771708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,3584,0.007959466675917308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,3584,0.01720213294029236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,4096,0.013690666357676188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,3072,0.007181866466999054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,3072,0.01723519961039225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,3584,0.012849066654841104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,2560,0.00631039987007777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,2560,0.017624533176422118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,3072,0.012020267049471537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,2048,0.005151999990145365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,2048,0.01685653328895569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,2560,0.011202133695284526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,1536,0.004463999966780345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,1536,0.014603733023007711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,2048,0.009991466999053955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,1536,0.00915839970111847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,1024,0.0034527999659379324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,1024,0.014707199732462563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,768,0.0032778667906920114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,768,0.015473066767056783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,768,0.007897600034872691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,512,0.0030239999294281008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,512,0.01599253316720327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,1024,0.00831573357184728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,256,0.002587733417749405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,256,0.014363732933998109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,256,0.007079466680685679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,128,0.0025173333783944448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,512,0.007506133119265239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,128,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,64,0.0021685334543387097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,64,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,4096,32,0.002569599946339925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,4096,32,0.014391466975212097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,128,0.007110400001207988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,65536,0.07458986441294352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,65536,0.07288320064544677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,16384,0.024311467011769613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,16384,0.032181332508722946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,5120,65536,0.12559786637624104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,12288,0.020097066958745323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,12288,0.028993066151936846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,16384,0.03581440051396688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,10240,0.01741973360379537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,10240,0.028493867317835493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,12288,0.028778666257858278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,8192,0.015305599570274353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,8192,0.023051732778549196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,8192,0.02108906706174215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,7168,0.013707733154296875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,7168,0.02218773365020752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,10240,0.024856533606847128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,6144,0.01402239998181661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,6144,0.02081386645634969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,7168,0.019410133361816406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,5120,0.010502400000890096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,5120,0.020332799355189005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,6144,0.017542399962743125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,4096,0.0086709330479304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,4096,0.01757546663284302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,5120,0.015701333681742348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,3584,0.007924266656239827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,3584,0.018117332458496095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,4096,0.01384106675783793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,3072,0.007041066884994507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,3072,0.01876266598701477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,4096,65536,0.1255946636199951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,2560,0.006405333181222279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,2560,0.01679253379503886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,3584,0.012845866878827415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,2048,0.005228800078233084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,2048,0.015936000148455302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,3072,0.012152533729871113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,1536,0.004366933306058248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,1536,0.017038933436075845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,1536,0.009248000383377076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,1024,0.003492266684770584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,1024,0.01461120049158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,2048,0.010048000017801921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,768,0.003387733300526937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,768,0.014507733782132468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,2560,0.010877866546312969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,512,0.003035733352104823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,512,0.015490133563677469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,1024,0.008384000261624653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,256,0.002749866743882497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,256,0.014526933431625366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,512,0.007652266820271809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,768,0.007909333209196727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,128,0.002619733413060506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,128,0.014499200383822122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,64,0.0021759999295075734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,64,0.0144405335187912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3584,32,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,128,0.006817066669464111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3584,32,0.014409599701563516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,65536,0.0671018679936727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,256,0.007119999825954437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,65536,0.06401386658350626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,16384,0.022014933824539184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,16384,0.02954453428586324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,12288,0.01727893352508545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,12288,0.02681066592534383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,12288,0.028169600168863933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,10240,0.015247999628384908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,10240,0.02483839988708496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,16384,0.036366931597391766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,8192,0.013229866822560629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,8192,0.022986666361490885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,10240,0.02481600046157837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,7168,0.011860266327857971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,7168,0.021370667219161987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,8192,0.021217066049575805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,6144,0.010994133353233338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,6144,0.02122453252474467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,7168,0.019093332688013713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,5120,0.009223467111587525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,5120,0.019348265727361043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,6144,0.017581866184870402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,4096,0.008239999910195668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,4096,0.017890133460362754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,5120,0.015737600127855935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,3584,0.007284266750017802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,3584,0.017452800273895265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,4096,0.01376106639703115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,3072,0.006836266815662384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,3072,0.016748799880345663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,3584,0.01286720037460327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,2560,0.005862399935722351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,2560,0.017148800690968833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3584,65536,0.12576853434244792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,2048,0.004822400212287903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,2048,0.016401066382726034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,3072,0.012120532989501952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,1536,0.004289066791534424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,1536,0.01594986617565155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,1536,0.00963200032711029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,1024,0.0035242666800816857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,1024,0.014983466267585755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,2560,0.010844799876213073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,768,0.0030613332986831666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,768,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,2048,0.010050132870674133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,512,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,512,0.014587733149528503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,768,0.007924266656239827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,256,0.0025621332228183747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,256,0.014413866400718688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,1024,0.008389332890510559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,512,0.00763626645008723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,128,0.002288000037272771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,128,0.014666666587193808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,64,0.002233600119749705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,64,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,3072,32,0.0022762666145960493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,128,0.007133866846561432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,256,0.007100800176461537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,3072,32,0.014105600118637086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,65536,0.05667946736017863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,16384,0.01996799906094869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,65536,0.06101119915644328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,16384,0.029090134302775066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,12288,0.015925332903862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,12288,0.02656426628430684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,12288,0.02885119915008545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,10240,0.014110933740933737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,10240,0.024774400393168132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,16384,0.035913598537445066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,8192,0.012264532844225566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,8192,0.022482132911682128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,10240,0.02506133317947388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,7168,0.011588266491889954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,7168,0.021676800648371377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,8192,0.02068480054537455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,6144,0.010414933164914448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,6144,0.0210591991742452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,7168,0.019246933857599895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,5120,0.009196799993515015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,5120,0.019806933403015137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,6144,0.017336533466974894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,4096,0.007760000228881836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,4096,0.018594133853912353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,5120,0.015828266739845276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,3584,0.007313066720962524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,3584,0.018708266814549766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,4096,0.013640532890955607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,3072,0.006322133541107178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,3072,0.016540799538294473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,3584,0.013318399588267008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,2560,0.005592533449331919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,2560,0.016670932372411094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,3072,0.011954133709271748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,2048,0.004635733366012573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,2048,0.01608746647834778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,2048,0.010315733154614766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,1536,0.0042346666256586705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,2560,0.011282133062680562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,1536,0.014501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,1024,0.0034282666941483817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,1024,0.014507733782132468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,1024,0.008326399823029835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,768,0.003036800026893616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,768,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,1536,0.00967039962609609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,512,0.0027104000250498454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,512,0.014618666966756186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,768,0.007968000074227651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,256,0.0028543998797734577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,256,0.014229333400726319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,512,0.007677866518497467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,128,0.0022698665658632917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,256,0.007083733379840851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,128,0.014593066771825156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,128,0.006684799989064534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,64,0.002170666555563609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,64,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2560,32,0.0022442666192849475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2560,32,0.014132266243298849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,65536,0.0466261347134908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,65536,0.05326186815897623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,3072,65536,0.12545173168182372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,16384,0.016302933295567833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,16384,0.028691200415293376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,12288,0.013539200027783712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,12288,0.025009065866470337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,12288,0.028166399399439497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,10240,0.012501333157221475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,10240,0.02417280077934265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,16384,0.035411198933919266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,8192,0.01107199986775716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,8192,0.025544534126917522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,10240,0.024778666098912557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,7168,0.010045866171518963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,7168,0.02158720095952352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,8192,0.020972800254821778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,6144,0.009533866246541341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,6144,0.021959465742111207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,7168,0.019539199272791543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,5120,0.008528000116348267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,5120,0.020759467283884683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,6144,0.017777067422866822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,4096,0.010588799913724262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,4096,0.01737066706021627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,5120,0.01574720044930776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,3584,0.009590400258700053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,3584,0.016620799899101257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,4096,0.013613866766293845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,3072,0.008707200487454731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,3072,0.016745599110921223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,3584,0.012862933675448099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,2560,0.006934399902820587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,2560,0.016520532965660095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,3072,0.011983999609947204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,2048,0.00472320020198822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,2048,0.017002665996551515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2560,65536,0.12559680143992108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,1536,0.0038890667259693147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,2560,0.011246933539708456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,1536,0.014518400033315023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,1024,0.0035477332770824433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,1024,0.014517333110173544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,2048,0.01002239982287089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,768,0.0030901332696278887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,768,0.014603733023007711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,1024,0.008302933474381765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,512,0.002666666607062022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,512,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,768,0.007999999821186066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,256,0.002586666742960612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,1536,0.00958720048268636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,256,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,128,0.0025279998779296877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,128,0.014568533500035605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,256,0.007165866593519847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,64,0.0022282667458057405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,128,0.006770133475462596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,512,0.00747626672188441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,2048,32,0.002570666621128718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,64,0.014169599612553915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,65536,0.03909226655960083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,2048,32,0.014439466595649719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,65536,0.047939201196034745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,16384,0.013867732882499696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,16384,0.02728319962819417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,12288,0.01151146690050761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,12288,0.025405865907669068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,12288,0.02802880009015401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,10240,0.010689066847165425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,10240,0.023534933725992836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,16384,0.03500906626383464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,8192,0.010447999835014344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,8192,0.021124267578125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,10240,0.024410667022069295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,7168,0.009681066870689392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,7168,0.020654932657877604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,8192,0.02072426676750183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,6144,0.009634133179982502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,6144,0.01911146640777588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,7168,0.019182932376861573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,5120,0.008405333757400513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,5120,0.019103999932607016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,6144,0.01735573410987854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,4096,0.007109333574771881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,4096,0.016927999258041383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,5120,0.01576746702194214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,3584,0.00629013329744339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,3584,0.017254400253295898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,4096,0.013649066289265951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,3072,0.0058783998092015585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,3072,0.017090133825937905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,3584,0.013237333297729493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,3072,0.011623467008272808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,2560,0.005046399931112925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,2560,0.016724266608556113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,2048,0.004752000172932943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,2048,0.016545066237449647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,2560,0.011187199751536052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,1536,0.003815466662247976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,1536,0.015878400206565856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,2048,0.009979733824729919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,1024,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,1536,0.009121066331863404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,1024,0.015553067127863566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,768,0.003009066730737686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,768,0.015641599893569946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,768,0.007977599898974102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,2048,65536,0.12486506303151448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,1024,0.008482133348782856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,512,0.0026517334083716077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,256,0.002644266684850057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,512,0.014490666985511779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,256,0.014215466380119324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,128,0.0022869333624839783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,128,0.014565333724021912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,256,0.007383466760317485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,64,0.002170666555563609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,512,0.007192533214886982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1536,32,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,64,0.014490666985511779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,128,0.006757333377997081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,65536,0.027889066934585573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1536,32,0.014230400323867798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,16384,0.010780800382296245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,65536,0.04331200122833252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,16384,0.027800534168879194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,12288,0.009412266810735067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,12288,0.024809600909550984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,12288,0.02765973409016927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,10240,0.009630933403968811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,10240,0.022870399554570518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,16384,0.03510719935099284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,8192,0.01341546674569448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,8192,0.02157333294550578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,10240,0.02438186605771383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,7168,0.010153599580128988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,7168,0.019352533419926963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,8192,0.020669867595036827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,6144,0.009634133179982502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,6144,0.01890346606572469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,7168,0.019014400243759156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,5120,0.00960533320903778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,5120,0.018733867009480796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,6144,0.01737173398335775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,4096,0.006774400174617767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,4096,0.0203167994817098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,5120,0.015416533748308817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,3584,0.006284800171852112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,3584,0.018662399053573607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,4096,0.013706666231155396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,3072,0.00544213354587555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,3072,0.01857173244158427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,3584,0.012898133198420206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,2560,0.005110399921735128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,2560,0.016633599996566772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,3072,0.012011733651161195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,2048,0.005793066819508871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,2048,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,2560,0.010804266730944315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,1536,0.005446400245030721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,1536,0.015848533312479655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,2048,0.010362666845321656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,1024,0.0040554667512575785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,1024,0.01446399986743927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,1536,0.009112532933553059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,768,0.0033802665770053864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,1024,0.008404266834259034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,768,0.014501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,512,0.003013333429892858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,512,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,768,0.007926400005817413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,256,0.0026133333643277483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,512,0.00747519979874293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,256,0.014498133460680643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,128,0.0024010665714740754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,128,0.014684800306955972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,256,0.007080533107121785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,64,0.002182399978240331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,64,0.014443733294804893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,128,0.007097599903742473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,1024,32,0.002346666653951009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,65536,0.02327893376350403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,1024,32,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,65536,0.03690239985783895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,16384,0.014619732896486918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,16384,0.026807467142740887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1536,65536,0.12444480260213216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,12288,0.012550399700800578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,12288,0.023600000143051147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,12288,0.027714133262634277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,10240,0.011542399724324543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,10240,0.023088000218073525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,16384,0.03465706507364909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,8192,0.010203733046849569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,8192,0.02087786595026652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,10240,0.02476480007171631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,7168,0.008389332890510559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,7168,0.019638399283091225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,8192,0.02073600093523661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,6144,0.007928533355395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,6144,0.01958079934120178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,7168,0.01899519960085551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,5120,0.007100800176461537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,5120,0.019423999389012656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,6144,0.017339734236399333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,4096,0.0066997334361076355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,4096,0.01829973260561625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,5120,0.015363199512163797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,3584,0.005940266450246175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,1024,65536,0.12230719725290935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,3584,0.016669867436091106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,3072,0.0054954667886098225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,3072,0.01653439998626709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,3072,0.011638399958610535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,3584,0.012426666418711345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,2560,0.00510506679614385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,4096,0.013333333532015481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,2560,0.016254933675130208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,2048,0.004350933432579041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,2048,0.0151829332113266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,1536,0.0038272000849246977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,1536,0.015999999642372132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,1536,0.009114666779836019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,1024,0.003399466723203659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,2048,0.010386133193969726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,2560,0.010790399710337321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,1024,0.014804266889890037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,768,0.003028266628583272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,768,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,512,0.0026047999660174055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,512,0.01443839967250824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,768,0.008066133161385854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,256,0.002609066665172577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,1024,0.008321066697438557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,512,0.007514666517575581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,256,0.013634133338928222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,128,0.0021749332547187803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,128,0.015044266978899637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,256,0.007144533097743988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,64,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,128,0.006726400057474773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,768,32,0.002183466653029124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,64,0.014437333742777506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,65536,0.017612799008687337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,768,32,0.013928533593813578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,16384,0.010752000411351522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,65536,0.03503146568934123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,16384,0.023411200443903605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,12288,0.009126399954160053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,12288,0.022395733992258707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,12288,0.0277834673722585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,10240,0.00844373305638631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,10240,0.021422932545344033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,16384,0.035207466284434004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,8192,0.007693866888682048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,8192,0.02262186606725057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,10240,0.024345600605010988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,7168,0.0070613334576288865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,7168,0.01928960084915161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,8192,0.020753065745035805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,6144,0.006781866649786632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,6144,0.019540266195933024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,7168,0.01900906761487325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,5120,0.006678399940331777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,5120,0.01815999945004781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,6144,0.017069866259892784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,4096,0.006868266562620799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,4096,0.019821866353352865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,5120,0.015689599514007568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,3584,0.005862399935722351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,3584,0.01801066597302755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,4096,0.013874133427937826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,3072,0.005541333556175232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,3072,0.01776319940884908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,768,65536,0.12128427028656005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,2560,0.005226666728655497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,3584,0.01288106640179952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,2560,0.01720959941546122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,2048,0.005182933310667674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,2048,0.016714666287104288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,3072,0.011691733201344808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,2560,0.01102186640103658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,1536,0.004282666742801667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,1536,0.015271466970443726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,1024,0.0034175999462604523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,1024,0.015185067057609558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,2048,0.010059733192125957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,768,0.0031541332602500914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,768,0.014556800325711569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,1536,0.009249066313107807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,1024,0.008306133250395458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,512,0.0026602665583292644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,512,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,768,0.008088533580303193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,512,0.007517866790294647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,256,0.0026837334036827086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,256,0.014172800381978354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,128,0.0022272000710169475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,128,0.014705066879590353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,256,0.007648000121116638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,64,0.002313599983851115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,128,0.006701866785685222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,512,32,0.0021930667261282604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,64,0.013677866260210673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,512,32,0.013763200243314108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,65536,0.016156799594561257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,65536,0.03356373310089111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,16384,0.007523199915885926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,16384,0.022666666905085245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,12288,0.007522133489449819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,12288,0.021312000354131062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,16384,0.03454506794611613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,10240,0.006234666705131531
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,10240,0.02141653299331665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,12288,0.02826559940973918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,8192,0.005995733539263407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,8192,0.020811732610066733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,10240,0.024297600984573363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,7168,0.005464533468087515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,7168,0.019215999046961467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,7168,0.018936532735824584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,6144,0.006257066627343495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,8192,0.02107200026512146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,6144,0.019662932554880778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,5120,0.006321066617965698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,5120,0.01870186726252238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,6144,0.017027199268341064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,5120,0.015702399611473083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,4096,0.005972266693909963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,4096,0.017905066410700478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,3584,0.0054624001185099285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,3584,0.016764799753824867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,4096,0.013658666610717773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,3584,0.012838400403658547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,3072,0.005508266886075338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,512,65536,0.12031466960906982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,3072,0.017409066359202065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,2560,0.005278933544953664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,2560,0.017529600858688356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,2048,0.004226133227348328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,2048,0.015025066335995993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,3072,0.012016000350316365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,1536,0.0037813333173592886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,2560,0.011244799693425496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,1536,0.0168938676516215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,2048,0.010388267040252686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,1024,0.0030613332986831666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,1024,0.014454399545987448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,768,0.0030037333567937215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,768,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,1536,0.009622400005658466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,1024,0.008397866288820903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,512,0.003014400104681651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,512,0.01589120030403137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,768,0.007905066510041555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,256,0.0025610665480295816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,512,0.00713919997215271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,128,0.0022069332500298817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,256,0.014406399925549826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,256,0.0070816000302632645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,128,0.014409599701563516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,64,0.0021898667017618816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,128,0.006733866532643636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,256,32,0.002216533323129018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,64,0.014514133334159851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,65536,0.011242666840553283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,256,32,0.01360106666882833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,16384,0.00588266650835673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,65536,0.03293866713841756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,16384,0.0231989324092865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,12288,0.005931733548641205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,12288,0.020837332804997763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,256,65536,0.12053120136260986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,12288,0.028012800216674804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,10240,0.006095999975999197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,16384,0.03458880186080933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,10240,0.020906666914621987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,8192,0.005569066603978475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,8192,0.019694934288660683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,7168,0.005444266895453135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,7168,0.018887466192245482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,8192,0.020771199464797975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,10240,0.024805333216985068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,6144,0.005909333129723867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,7168,0.018971733252207437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,5120,0.005875200033187866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,6144,0.018869332472483315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,5120,0.019191465775171914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,4096,0.005444266895453135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,4096,0.018539732694625853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,6144,0.017343999942143758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,3584,0.005098666747411092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,5120,0.015370666980743408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,3584,0.016676266988118492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,4096,0.013664000233014426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,3072,0.0050570666790008545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,3072,0.01658453345298767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,3584,0.012936533490816752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,2560,0.00537066658337911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,3072,0.011993599931399028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,2560,0.017127466201782227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,2048,0.005138133466243744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,2560,0.0107914666334788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,2048,0.016562133034070333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,1536,0.0038783999780813852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,2048,0.01002239982287089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,1536,0.014984533190727234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,1024,0.0031530665854612983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,1536,0.009165866176287334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,1024,0.014630400141080222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,768,0.003094399968783061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,1024,0.008368000388145447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,768,0.014717866977055868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,768,0.00791786660750707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,512,0.0026528000831604003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,512,0.013414399822553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,65536,0.11993066469828288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,256,0.0022314667701721193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,512,0.007573333382606506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,128,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,256,0.00747519979874293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,256,0.0146506667137146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,64,0.002268799891074498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,128,0.01437333325544993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,128,32,0.002293333411216736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,80,128,128,0.006678399940331777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,64,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,65536,0.008343467116355896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,128,32,0.01439786652723948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,16384,0.005907199780146281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,12288,0.005458133419354757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,65536,0.02873706618944804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,16384,0.021755733092625937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,10240,0.005982933441797892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,12288,0.020807466904322305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,8192,0.005524266759554545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,10240,0.020662399133046468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,7168,0.005463466544946035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,8192,0.019051732619603474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,6144,0.005094400048255921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,5120,0.004692266881465912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,4096,0.004622933268547058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,6144,0.018658133347829182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,7168,0.020746666193008422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,5120,0.018992000818252565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,3584,0.0047199999292691554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,3072,0.004649599889914194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,4096,0.018307199080785118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,2560,0.005127466718355815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,2048,0.005065600077311197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,3072,0.016754132509231568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,3584,0.01664959987004598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,1536,0.0037685332198937735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,2560,0.01599253316720327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,1024,0.0032373333970705668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,2048,0.015897599856058757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,1536,0.01480959951877594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,768,0.0030250666042168934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,512,0.0025685332715511323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,1024,0.01575040022532145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,256,0.0021749332547187803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,768,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,512,0.014455466469128927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,128,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,64,0.0021375998854637145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,256,0.013501866658528646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,128,0.014262400070826211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,64,32,0.0022101332743962605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,64,0.013294933239618936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,65536,0.009196799993515015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,16384,0.006257066627343495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,64,32,0.013306666413942972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,65536,0.02879040042559306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,12288,0.005093333125114441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,16384,0.02176106572151184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,10240,0.00544106662273407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,8192,0.005486933390299479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,12288,0.02067520022392273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,10240,0.020528000593185425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,7168,0.005176533261934916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,8192,0.018850133816401164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,6144,0.005010133484999338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,5120,0.0046847999095916745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,7168,0.018971733252207437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,6144,0.019591466585795084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,4096,0.004588800172011057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,5120,0.01862506667772929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,3584,0.004715733230113983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,3072,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,4096,0.01849493384361267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,2560,0.005041066805521647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,3584,0.017003732919692992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,3072,0.016548267006874083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,2048,0.004681600133577982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,1536,0.0033887999753157297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,2560,0.01664426624774933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,2048,0.016549332936604818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,1024,0.002974933385848999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,768,0.0026602665583292644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,1536,0.014491732915242514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,512,0.0025813333690166474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,1024,0.014514133334159851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,768,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,256,0.0022069332500298817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,128,0.002218666672706604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,512,0.014629333217938741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,256,0.013013333082199097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,64,0.0018272000054518384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,80,32,32,0.0019061333189407986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,128,0.0138154665629069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,64,0.014095999797185264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,80,32,32,0.014312533537546792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,16384,0.2052992026011149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,16384,0.37554667790730795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,12288,0.2688245455423991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,12288,0.15793066024780272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,10240,0.22893013954162597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,10240,0.1305450677871704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,8192,0.18638720512390136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,8192,0.11223893165588379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,8192,0.06228906710942587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,7168,0.15467413266499835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,7168,0.0957749366760254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,10240,0.07322239875793457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,6144,0.14302293459574383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,6144,0.08596479892730713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,12288,0.08781546751658122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,5120,0.11426133314768475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,5120,0.07320746580759684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,6144,0.04755733410517375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,4096,0.10069973468780517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,4096,0.06542719999949137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,7168,0.054390398661295566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,3584,0.0804149309794108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,3584,0.057765332857767734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,5120,0.04214186668395996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,3072,0.07166826725006104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,3072,0.05196373462677002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,16384,0.11591146787007649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,2560,0.05957653522491455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,2560,0.047984000047047934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,3584,0.03191039959589641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,2048,0.048951466878255204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,2048,0.04122666517893474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,4096,0.03595413366953532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,1536,0.03767040173212687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,1536,0.033801599343617754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,3072,0.030407466491063434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,1024,0.02807786663373311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,1024,0.026786132653554277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,2560,0.026451200246810913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,768,0.021246933937072755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,768,0.024759467442830405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,2048,0.02202026645342509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,512,0.01411626636981964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,512,0.021283199389775596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,1536,0.01775040030479431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,256,0.008447999755541485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,256,0.01660053332646688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,768,0.01290880044301351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,128,0.004826666911443075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,128,0.015601066748301187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,1024,0.014298666516939798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,64,0.003841066608826319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,64,0.016479999820391337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,65536,32,0.004249600072701773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,65536,32,0.015126400192578635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,128,0.008372267087300617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,512,0.010851200421651204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,65536,0.3417130788167318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,65536,0.21306239763895668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,16384,0.08630080223083496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,16384,0.06080960035324097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,12288,0.06146453221638998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,12288,0.049667199452718094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,65536,256,0.009463466207186381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,10240,0.05318933327992757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,10240,0.04422826766967773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,10240,0.028586665789286297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,8192,0.04321493307749431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,8192,0.03732586701711019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,12288,0.032553599278132124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,7168,0.03720959822336833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,7168,0.034986666838328045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,16384,0.042011733849843344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,6144,0.033321599165598556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,6144,0.03183466593424479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,8192,0.023950932423273723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,5120,0.028376533587773638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,5120,0.02895680069923401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,7168,0.021512534221013388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,4096,0.022768000761667885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,4096,0.026739199956258137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,5120,0.016956800222396852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,3584,0.02031893332799276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,3584,0.025044266382853193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,6144,0.0192522664864858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,3072,0.017937066157658894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,3072,0.02290133237838745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,4096,0.015075199802716575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,2560,0.015500799814860026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,2560,0.021350399653116862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,3584,0.01416106621424357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,2048,0.01248426636060079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,2048,0.021910399198532104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,2560,0.012162133057912191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,1536,0.010059733192125957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,1536,0.018798933426539103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,3072,0.013352533181508383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,1024,0.007518933216730754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,1024,0.014797866344451904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,1536,0.009751466910044353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,768,0.006371200084686279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,768,0.0164874662955602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,1024,0.008343467116355896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,512,0.003916800022125244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,512,0.015656532843907674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,512,0.007537066439787547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,256,0.003409066547950109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,256,0.01455466647942861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,2048,0.010925867160161336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,128,0.003156266609827677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,128,0.014769066373507181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,768,0.00801386684179306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,64,0.0030464000999927522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,64,0.014532267053922018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,16384,32,0.0030901332696278887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,16384,32,0.014569600423177084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,65536,0.24701333045959473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,128,0.007155199845631917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,256,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,65536,0.16051626205444336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,16384,0.07005226612091064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,16384,0.05085119803746542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,12288,0.04718720118204753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,12288,0.04194986820220947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,12288,0.030869332949320476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,10240,0.04015466769536336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,10240,0.037419732411702475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,16384,0.03914560079574585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,8192,0.03369599978129069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,8192,0.0327349325021108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,10240,0.027030400435129803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,7168,0.02898453275362651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,7168,0.030522666374842328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,8192,0.022331732511520385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,6144,0.02597973346710205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,6144,0.02785173257191976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,7168,0.02026346723238627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,5120,0.022029866774876915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,5120,0.024963200092315674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,6144,0.01835093299547831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,4096,0.017905066410700478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,4096,0.02487893303235372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,5120,0.01683626572291056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,3584,0.01585599978764852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,3584,0.022695465882619222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,4096,0.014773333072662353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,3072,0.014550399780273438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,3072,0.022810665766398112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,3584,0.013777066270510355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,2560,0.012560000022252401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,2560,0.020833067099253335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,3072,0.012470400333404541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,2048,0.010850133498509724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,2048,0.020220800240834554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,2560,0.011648000280062357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,1536,0.00897173285484314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,1536,0.01846826672554016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,1536,0.009266133109728496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,1024,0.006750933329264323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,1024,0.01753386656443278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,2048,0.010063999891281128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,768,0.0051125332713127134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,768,0.014750933647155762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,768,0.007899733384450276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,512,0.0038143999874591826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,512,0.015779200196266174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,1024,0.008408533533414204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,256,0.0032490665713946023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,256,0.014550399780273438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,256,0.007239466905593872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,128,0.00308693324526151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,128,0.014571733276049294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,512,0.007152000069618225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,64,0.002619733413060506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,64,0.014541866381963095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,12288,32,0.002993066608905792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,12288,32,0.014502400159835815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,65536,0.21378347078959145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,128,0.007149866720040639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,65536,0.1445184071858724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,16384,0.05533759991327921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,16384,0.04542719920476278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,16384,0.03793280124664307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,12288,0.03985386689503988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,12288,0.037855998675028486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,12288,0.02980479995409648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,10240,0.034086398283640545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,10240,0.033463466167449954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,16384,65536,0.144269863764445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,8192,0.02834666570027669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,8192,0.030158933003743487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,10240,0.02595733404159546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,7168,0.024553600947062174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,7168,0.026964267094930012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,12288,65536,0.13268160025278727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,6144,0.021913599967956544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,6144,0.02508266568183899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,7168,0.02028800050417582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,5120,0.018910932540893554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,5120,0.02328746716181437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,8192,0.022105600436528525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,4096,0.01532799998919169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,4096,0.02320746580759684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,4096,0.0149536003669103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,3584,0.01339413324991862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,3584,0.02141973376274109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,5120,0.01660586694876353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,3072,0.011764267086982727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,3072,0.020829866329828896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,6144,0.01874026656150818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,2560,0.010455466310183207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,2560,0.02048106590906779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,3072,0.012087466319402058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,2048,0.008778666456540424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,2048,0.01727786660194397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,3584,0.013910399874051413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,1536,0.007388799885908763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,1536,0.01660480002562205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,2560,0.011480533083279927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,1024,0.0052042668064435325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,1024,0.015549866358439126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,1024,0.008469333251317341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,768,0.003984000037113826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,768,0.015904000401496886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,1536,0.009742933511734008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,512,0.0034741332133611047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,512,0.014789332946141561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,768,0.008408533533414204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,256,0.0030794667700926462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,256,0.015177599589029946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,512,0.007650133470694225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,128,0.0027114666998386385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,2048,0.010389332969983418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,128,0.014615466197331747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,64,0.0027146667242050173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,64,0.014710399508476257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,10240,32,0.0026752000053723653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,10240,32,0.01479680041472117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,65536,0.1800938606262207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,128,0.006664533416430156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,256,0.007153066496054332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,65536,0.12065172990163167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,16384,0.04695146481196086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,16384,0.04100586573282878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,12288,0.03182613253593445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,12288,0.03450133403142293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,12288,0.02901653250058492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,10240,0.02750399907430013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,10240,0.030819199482599896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,16384,0.03718506495157878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,8192,0.022367999951044718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,8192,0.026447999477386474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,10240,0.025623466571172076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,7168,0.020011732975641884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,7168,0.026681600014368693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,8192,0.021874133745829263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,6144,0.017787732680638633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,6144,0.024546132485071818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,7168,0.02029119928677877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,5120,0.015159466862678527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,5120,0.02295680046081543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,10240,65536,0.1287712017695109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,4096,0.0121888001759847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,4096,0.02219520012537638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,6144,0.018231467405954997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,3584,0.01221226652463277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,3584,0.02063039938608805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,5120,0.016671999295552572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,3072,0.011005866527557372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,3072,0.020551466941833497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,4096,0.014065066973368326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,2560,0.009604266285896302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,2560,0.01876586675643921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,3584,0.013304533561070761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,2048,0.008268799881140392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,2048,0.01811199982961019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,2560,0.010789333780606588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,1536,0.006705066561698914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,1536,0.016151466965675355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,2048,0.01001706620057424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,3072,0.012418133020401002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,1024,0.004957866668701172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,1024,0.014753066500027976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,768,0.0038506666819254553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,768,0.014846932888031007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,1536,0.009191466371218364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,512,0.0030559999247392017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,512,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,1024,0.008707200487454731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,256,0.0029386666913827257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,768,0.00793280005455017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,256,0.015431466698646545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,128,0.002961066613594691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,128,0.014971733093261719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,128,0.006763733426729838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,64,0.0027072000006834666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,64,0.01460693379243215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,8192,32,0.002977066735426585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,8192,32,0.014712533354759217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,256,0.007449600100517273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,65536,0.15706027348836263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,512,0.007528533538182576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,65536,0.1057535966237386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,16384,0.04585280021031697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,16384,0.04233173529307048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,12288,0.03518079916636149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,12288,0.03243733247121175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,12288,0.02890026569366455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,10240,0.026459733645121258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,10240,0.02897599935531616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,16384,0.036297599474589035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,8192,0.021653334299723305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,8192,0.025493333737055462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,10240,0.0256117324034373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,7168,0.018898133436838785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,7168,0.024716800451278685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,8192,0.021869866053263347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,6144,0.0167413334051768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,6144,0.02476693391799927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,7168,0.01986560026804606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,5120,0.014494933684666953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,5120,0.02270080049832662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,6144,0.018196266889572144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,4096,0.011608533064524333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,4096,0.021107200781504312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,5120,0.01634880006313324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,3584,0.010642133156458537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,3584,0.020651733875274657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,4096,0.013957333564758301
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,3072,0.010321066776911417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,3072,0.021398399273554483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,3584,0.013268267114957174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,2560,0.009015466769536335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,2560,0.017435733477274576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,3072,0.011623467008272808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,2048,0.007898666461308797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,2048,0.016535466909408568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,2560,0.010823466380437215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,1536,0.006391466657320659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,1536,0.015082666277885437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,1536,0.00959999958674113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,1024,0.004257066547870636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,1024,0.015281066298484802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,2048,0.010468266407648722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,768,0.003537066777547201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,768,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,1024,0.008389332890510559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,512,0.003048533449570338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,512,0.015411200126012168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,768,0.007921066880226136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,256,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,256,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,512,0.007531733314196269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,128,0.002619733413060506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,128,0.014705066879590353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,256,0.007092266778151194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,64,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,64,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,7168,32,0.0026357332865397137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,128,0.007150933146476746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,7168,32,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,65536,0.1226687987645467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,65536,0.09039786656697592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,16384,0.03882879813512166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,16384,0.040226133664449056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,8192,65536,0.1264074643452962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,12288,0.030557866891225176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,12288,0.034024532636006674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,16384,0.03600746790568034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,10240,0.02666986584663391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,10240,0.03299840092658997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,12288,0.028779733180999755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,8192,0.024190932512283325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,8192,0.025413332382837932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,10240,0.025553067525227863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,7168,0.02157013416290283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,7168,0.02397759954134623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,8192,0.021475199858347574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,6144,0.015336533387502035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,6144,0.022500266631444298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,7168,0.02018346587816874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,5120,0.013312000036239623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,5120,0.02113706668217977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,6144,0.017821866273880004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,4096,0.01115733285744985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,4096,0.021171200275421142
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,5120,0.01616853376229604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,3584,0.01016426682472229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,4096,0.013736533125241599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,3584,0.02013546625773112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,3072,0.008706133564313252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,3072,0.017448532581329345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,3072,0.012151466806729634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,2560,0.0076330666740735366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,2560,0.01713493267695109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,3584,0.012836266557375589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,2048,0.006691200037797292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,2048,0.015446399648984274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,7168,65536,0.1261141300201416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,1536,0.005151999990145365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,1536,0.015896532932917276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,2048,0.009963732957839967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,1024,0.003835733234882355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,1024,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,2560,0.010972799857457478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,768,0.0035189333061377203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,768,0.014692266782124838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,1024,0.008321066697438557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,768,0.008071466783682505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,512,0.002979200085004171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,512,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,256,0.0026933332284291584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,256,0.014680533607800802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,1536,0.009172266721725464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,128,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,128,0.014647466937700906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,256,0.0071168000499407455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,64,0.002271999915440877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,128,0.0070165331164995836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,6144,32,0.0025557334224383037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,64,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,6144,32,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,65536,0.10483307043711346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,512,0.007522133489449819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,65536,0.08541119893391927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,16384,0.033078400293986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,16384,0.037384533882141115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,12288,0.025860265890757246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,12288,0.031717334191004434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,12288,0.02850773334503174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,10240,0.022778666019439696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,10240,0.030763733386993408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,16384,0.035795199871063235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,8192,0.018992000818252565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,8192,0.024408533175786336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,10240,0.025547732909520466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,7168,0.018811732530593872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,7168,0.02324906587600708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,8192,0.021547732750574748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,6144,0.016988799969355265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,6144,0.02213866710662842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,7168,0.019951999187469482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,5120,0.012307199835777282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,5120,0.020948266983032225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,6144,0.017426133155822754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,4096,0.010024533669153849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,4096,0.01870186726252238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,5120,0.015734400351842245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,3584,0.009402666489283245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,3584,0.01922453244527181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,4096,0.013779200116793313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,3072,0.008344533046086629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,3072,0.016661333044370015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,3584,0.012964266538619994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,2560,0.007272533575693766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,2560,0.017896533012390137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,3072,0.01200213332970937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,2048,0.005762133498986562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,2048,0.014921599626541137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,2560,0.010967466235160827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,1536,0.00476800004641215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,1536,0.016474666198094685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,2048,0.00999679962793986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,1024,0.003841066608826319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,1024,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,1536,0.009388800462086995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,768,0.003483733286460241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,768,0.01632960041364034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,1024,0.008316799998283386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,512,0.0030037333567937215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,512,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,6144,65536,0.126419202486674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,256,0.0027722666660944624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,256,0.014070399602254233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,512,0.007521066566308339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,128,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,128,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,256,0.007590400179227193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,64,0.0022677332162857056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,64,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,768,0.008032000064849854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,128,0.007085866729418437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,5120,32,0.002628266563018163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,5120,32,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,65536,0.09105280240376791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,65536,0.07365439732869467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,16384,0.025909332434336345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,16384,0.03130346735318502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,12288,0.021383466323216756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,12288,0.027726932366689043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,12288,0.028812799851099653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,10240,0.017465599377950034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,10240,0.025110399723052977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,16384,0.03589119911193848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,8192,0.014663466811180114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,8192,0.023902932802836098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,10240,0.025608533620834352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,7168,0.013091199596722922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,7168,0.02158400019009908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,8192,0.021090133984883627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,6144,0.01160533328851064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,6144,0.020653865734736123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,7168,0.019820799430211387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,5120,0.010339200496673584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,5120,0.01880000034968058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,6144,0.017372800906499227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,4096,0.008776533603668212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,4096,0.018819200992584228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,5120,0.016083199779192606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,3584,0.008090666433175405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,3584,0.016823466618855795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,4096,0.013358933726946512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,3072,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,3072,0.016782933473587038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,3072,0.011708799997965496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,2560,0.006056533257166544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,2560,0.01646613379319509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,3584,0.012980266412099203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,2048,0.005186133086681366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,2048,0.016341333587964378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,2560,0.01081706682840983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,1536,0.0038783999780813852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,1536,0.018256000677744546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,2048,0.010578133662541707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,1024,0.0036277333895365395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,1024,0.014659200112024942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,1024,0.008452266454696655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,768,0.0031487998863061273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,768,0.014731733004252115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,1536,0.009147733449935913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,512,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,512,0.014525866508483887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,512,0.007114666700363159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,256,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,768,0.008070399860541026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,256,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,128,0.002152533332506816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,128,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,128,0.006691200037797292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,64,0.002165333429972331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,64,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,256,0.007560533285140991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,4096,32,0.002229333420594533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,4096,32,0.01458026667435964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,65536,0.07553706963857015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,65536,0.06504000027974446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,16384,0.02442453304926554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,16384,0.03126186728477478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,5120,65536,0.1255893309911092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,12288,0.019182932376861573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,12288,0.02664533257484436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,16384,0.03619199991226196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,10240,0.01616320013999939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,10240,0.024053333202997844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,12288,0.02847359975179036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,8192,0.014966400464375815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,8192,0.022216532627741495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,8192,0.02095573345820109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,7168,0.013556266824404398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,7168,0.020911999543507895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,10240,0.02566186587015788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,6144,0.01125440001487732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,6144,0.02116159995396932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,7168,0.01941546599070231
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,5120,0.009846400221188862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,5120,0.01857066750526428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,6144,0.01737066706021627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,4096,0.008389332890510559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,4096,0.01746986707051595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,4096,0.013680000106493631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,3584,0.0075434664885203045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,3584,0.017717333634694417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,5120,0.015752533078193666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,3072,0.006842666864395141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,3072,0.017059199015299478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,3584,0.012888532876968384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,2560,0.005646933118502299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,2560,0.016574933131535848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,3072,0.012014933427174886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,2048,0.0046623999873797095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,4096,65536,0.1253706693649292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,2048,0.01620266636212667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,1536,0.004011733333269755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,1536,0.01586133340994517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,2560,0.011245866616566975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,1024,0.0034453332424163817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,1024,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,2048,0.010385066270828247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,768,0.003142400085926056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,768,0.014542933305104574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,1536,0.009142399827639262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,512,0.002757333219051361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,512,0.01562773287296295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,768,0.00790293316046397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,256,0.0025642665723959604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,256,0.014573867122332254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,1024,0.008222933113574981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,128,0.0021759999295075734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,128,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,512,0.007162666817506154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,64,0.0022282667458057405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,64,0.014494933684666953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3584,32,0.002600533266862233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,256,0.007482666770617168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3584,32,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,128,0.0070698668559392285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,65536,0.066594131787618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,16384,0.02323733369509379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,65536,0.061020799477895106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,16384,0.02990293304125468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,12288,0.017780266205469766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,12288,0.028333866596221925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,12288,0.028523733218510944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,10240,0.0161461333433787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,10240,0.023587199052174886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,16384,0.03585066795349121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,8192,0.013637333114941915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,8192,0.022779732942581177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,10240,0.025131734212239583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,7168,0.012460800011952718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,7168,0.021117866039276123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,8192,0.02102400064468384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,6144,0.011479467153549194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,6144,0.018807466824849448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,7168,0.019433599710464478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,5120,0.009352533022562663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,5120,0.01984106699625651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,6144,0.017369600137074788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,4096,0.007942399879296621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,4096,0.017296000321706136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,5120,0.015828266739845276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,3584,0.007229866584142049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,3584,0.017369600137074788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,4096,0.01369599997997284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,3072,0.00631466656923294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,3072,0.016777600844701132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,3584,0.012924800316492716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,2560,0.005514666438102722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,2560,0.016463999946912132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,3072,0.011783466736475626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3584,65536,0.1252405325571696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,2048,0.004855466882387797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,2048,0.016663466890652977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,1536,0.004227200150489807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,1536,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,2560,0.011210667093594869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,1024,0.0035146666069825493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,1024,0.014979199568430582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,1536,0.009157333771387737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,768,0.003070933371782303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,768,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,1024,0.008469333251317341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,512,0.0030720000465710956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,512,0.01464959979057312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,2048,0.010045866171518963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,256,0.002696533252795537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,256,0.015096533298492431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,768,0.007952000200748443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,128,0.0025642665723959604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,128,0.014547200004259745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,256,0.007161599894364675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,64,0.0022783999641736346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,512,0.007604266703128815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,3072,32,0.0025770666698614756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,128,0.006690133114655812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,64,0.014588800072669984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,3072,32,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,65536,0.0570581316947937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,16384,0.01849493384361267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,65536,0.0572490652402242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,16384,0.028886399666468304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,12288,0.01683626572291056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,12288,0.02611839969952901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,12288,0.028381866216659547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,10240,0.01498240033785502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,10240,0.02313813368479411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,16384,0.035529601573944095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,8192,0.012871467073758445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,8192,0.021112533410390218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,10240,0.02478933334350586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,7168,0.012242133418718975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,7168,0.020627200603485107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,8192,0.020778665939966835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,6144,0.01114453375339508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,6144,0.01909653345743815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,7168,0.019077332814534505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,5120,0.009223467111587525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,5120,0.01872746745745341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,6144,0.01778986652692159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,4096,0.007796266674995422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,4096,0.01684373418490092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,5120,0.01576640009880066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,3584,0.006695466736952464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,3584,0.017112533251444496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,4096,0.013784533739089966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,3072,0.00606826643149058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,3072,0.016667733589808144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,3584,0.01283519963423411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,2560,0.0050805335243542995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,2560,0.01566933294137319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,3072,0.012199466427167256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,2048,0.004710400104522705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,2048,0.014888532956441245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,2560,0.010827733079592387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,1536,0.0042250668009122215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,1536,0.015715199708938598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,2048,0.010126933455467224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,1024,0.003575466573238373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,1024,0.014598400394121806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,1536,0.009113599856694538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,768,0.0030239999294281008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,768,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,3072,65536,0.1257525364557902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,768,0.007935999830563863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,512,0.003202133377393087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,512,0.015874133507410685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,256,0.0026357332865397137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,1024,0.008480000495910644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,256,0.014497066537539164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,128,0.0021674667795499166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,128,0.01440000037352244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,128,0.006793599824110668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,256,0.007160533467928569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,64,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2560,32,0.002419200042883555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,64,0.014522666732470194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2560,32,0.014376533031463624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,512,0.007965866724650066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,65536,0.047033600012461346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,16384,0.017307732502619425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,65536,0.049991468588511154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,16384,0.02684906721115112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,12288,0.013671466708183288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,12288,0.024563199281692503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,12288,0.027811199426651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,10240,0.01232319970925649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,10240,0.022758400440216063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,16384,0.03545173406600952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,8192,0.010759466886520385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,8192,0.022267733017603555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,10240,0.02446720004081726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,7168,0.010053333640098572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,7168,0.019758933782577516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,8192,0.02063466707865397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,6144,0.009187199672063192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,6144,0.01927786668141683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,7168,0.019182932376861573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,5120,0.008161066472530365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,5120,0.019219199816385903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,6144,0.017343999942143758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,4096,0.00717439999183019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,4096,0.016759467124938966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,4096,0.01334826648235321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,3584,0.006301866471767425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,5120,0.015982932845751443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,3584,0.016730666160583496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,3072,0.005465599894523621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,3072,0.016775466998418174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,3072,0.011598933736483257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,2560,0.005047466854254405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,2560,0.01576746702194214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,3584,0.01262399951616923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,2048,0.004763733347256978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,2048,0.014908799529075622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,2560,0.010802132884661357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,1536,0.0038250667353471124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,1536,0.014732799927393594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,2048,0.010073600212732951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,1024,0.003492266684770584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,1024,0.014605866869290671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,1024,0.008371200164159138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,1536,0.009142399827639262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,768,0.002979200085004171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2560,65536,0.12527680397033691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,512,0.0030762667457262674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,768,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,512,0.014562132954597472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,256,0.002677333354949951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,256,0.015537066260973611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,512,0.0071263998746871945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,256,0.00714026689529419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,128,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,768,0.007880533238252004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,128,0.014904533823331198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,64,0.0021898667017618816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,128,0.0067093332608540845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,64,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,2048,32,0.00225600004196167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,65536,0.03550826708475749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,2048,32,0.014428800344467163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,65536,0.04490986665089925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,16384,0.013723733027776084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,16384,0.02558079957962036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,12288,0.011326932907104492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,12288,0.023510400454203287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,12288,0.027720532814661664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,10240,0.010411733388900756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,10240,0.022021333376566567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,16384,0.03549439907073974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,8192,0.009455999732017517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,8192,0.02132373253504435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,10240,0.02440213362375895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,7168,0.008425600330034892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,7168,0.01950613260269165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,8192,0.02078933318456014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,6144,0.007702399790287018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,6144,0.019401599963506065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,7168,0.01902186671892802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,5120,0.007463466624418895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,5120,0.018540799617767334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,6144,0.01718506614367167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,4096,0.006860800087451935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,4096,0.01676266590754191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,5120,0.015205333630243937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,3584,0.005870933334032694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,3584,0.016720000902811685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,4096,0.01386666695276896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,3072,0.00556160012880961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,3072,0.016727467377980552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,3584,0.012402133146921793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,2560,0.0050335998336474095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,2560,0.016323199868202208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,3072,0.011700266599655151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,2048,0.004750933249791463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,2048,0.016251732905705772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,2560,0.010788266857465107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,1536,0.0038751999537150065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,1536,0.014603733023007711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,2048,0.009666132926940917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,1024,0.0035573333501815797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,1536,0.008796800176302593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,1024,0.014604799946149192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,768,0.0030613332986831666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,768,0.014562132954597472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,2048,65536,0.1253397305806478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,512,0.002762666592995326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,1024,0.00790293316046397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,512,0.01567466656366984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,256,0.002630399912595749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,768,0.007525333265463512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,256,0.013578666249910989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,128,0.0022240000466505687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,128,0.014072533448537192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,512,0.007534933090209961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,256,0.007176533341407776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,64,0.0021685334543387097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1536,32,0.0021397332350413007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,64,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1536,32,0.013392000397046407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,65536,0.029317333300908403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,128,0.006730666756629944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,16384,0.011013333002726238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,65536,0.03903146584828694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,16384,0.022900267442067464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,12288,0.009318400422732036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,12288,0.021927465995152794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,12288,0.02815999984741211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,10240,0.008184533317883809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,10240,0.021437867482503255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,16384,0.03507733345031738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,8192,0.0076223999261856076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,8192,0.020528000593185425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,10240,0.024471465746561685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,7168,0.006863999863465626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,7168,0.01946880022684733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,8192,0.02067626714706421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,6144,0.0063285330931345625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,6144,0.019142399231592812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,6144,0.017323732376098633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,5120,0.0066655998428662615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,5120,0.01807360053062439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,7168,0.01911679903666178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,4096,0.006056533257166544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,4096,0.016942934195200602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,5120,0.015282133221626281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,3584,0.005927466849486033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,3584,0.016572800278663636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,4096,0.013413332899411521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,3072,0.005622399846712748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,3072,0.01678826610247294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,3584,0.012410666545232136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,2560,0.0055167997876803074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,2560,0.017273600896199545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,3072,0.011610666910807293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,2048,0.004242133100827535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,2048,0.015499732891718545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,2560,0.011325866977373759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,1536,0.00393599991997083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,2048,0.009664000074068705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,1536,0.014923733472824097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,1024,0.003172266731659571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,1024,0.015936000148455302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,1536,0.009125333031018574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,768,0.002930133293072383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,1024,0.00867306689421336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,768,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,512,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,512,0.01477013329664866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,512,0.007520000139872233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,768,0.008035199840863545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,256,0.002567466596762339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,256,0.014501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,128,0.002170666555563609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1536,65536,0.12455573081970214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,256,0.007096533477306366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,64,0.0021503999829292296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,128,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,1024,32,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,64,0.014086400469144186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,128,0.0071712002158164975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,1024,32,0.01442346672217051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,65536,0.02347946763038635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,16384,0.00960533320903778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,65536,0.035146665573120114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,16384,0.02236586610476176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,12288,0.007618133227030437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,12288,0.021623466412226358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,12288,0.027592533826828004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,10240,0.0070826664566993715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,10240,0.02181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,16384,0.03524800141652425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,8192,0.00647573322057724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,8192,0.020758400360743202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,10240,0.023935999472935995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,7168,0.005966933568318685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,7168,0.02001706759134928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,8192,0.020716800292332967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,6144,0.006470400094985962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,6144,0.020514132579167683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,7168,0.01892906626065572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,5120,0.006676266590754191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,5120,0.01878933310508728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,6144,0.017465599377950034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,4096,0.006417066852251689
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,4096,0.01691306630770365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,5120,0.015310933192571005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,3584,0.005952000121275584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,3584,0.016503467162450155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,1024,65536,0.12180053393046061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,3072,0.005681066711743673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,3072,0.01755519906679789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,3584,0.012446932991345723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,2560,0.005471999943256378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,2560,0.016520532965660095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,4096,0.013666133085886637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,2048,0.004484266539414724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,2048,0.016013866662979125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,3072,0.011752532919247945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,1536,0.0039488000174363455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,1536,0.016327466567357382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,2560,0.01076800028483073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,1024,0.0030591999491055804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,1024,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,2048,0.01011840005715688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,768,0.003101866692304611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,768,0.014696533481280008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,1024,0.007960533102353413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,512,0.002566399921973546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,1536,0.00922986666361491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,768,0.008041599889596303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,256,0.0026389333109060925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,512,0.015141333142916361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,256,0.014774399995803832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,128,0.002327466756105423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,256,0.007092266778151194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,512,0.007339733342329661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,128,0.014661332964897156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,64,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,768,32,0.0022410665949185687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,64,0.014502400159835815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,128,0.006694399813810985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,768,32,0.013346133629480996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,65536,0.01739413340886434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,65536,0.03092479904492696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,16384,0.007005866865317028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,16384,0.022051199277242025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,12288,0.00665280024210612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,12288,0.02198186715443929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,12288,0.027640533447265626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,10240,0.00622506688038508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,16384,0.03504106601079305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,10240,0.02186773419380188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,8192,0.006084266801675161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,8192,0.02069013317426046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,10240,0.0243776003519694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,7168,0.00610453337430954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,7168,0.019483733177185058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,8192,0.020691200097401937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,6144,0.005905066430568695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,6144,0.01913599967956543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,7168,0.019062399864196777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,5120,0.006807466844717662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,5120,0.0189194659392039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,6144,0.017336533466974894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,4096,0.0058559998869895935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,4096,0.01912320057551066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,5120,0.015429332852363586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,3584,0.005977599819501241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,4096,0.01362666686375936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,3584,0.017033600807189943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,3072,0.005497600138187409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,3072,0.01681813398996989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,3584,0.012423466642697651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,2560,0.005066666503747304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,2560,0.015313067038853965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,3072,0.011735467116038005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,2048,0.005117866893609365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,2048,0.016432000199953715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,768,65536,0.1208682696024577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,1536,0.003955200066169103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,1536,0.01536853313446045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,2560,0.010792533556620281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,1024,0.0030645333230495454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,1024,0.014491732915242514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,2048,0.009956266482671101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,768,0.002962133288383484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,768,0.01460693379243215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,1536,0.009227733810742695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,1024,0.007986133297284443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,512,0.0026848000784715016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,512,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,256,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,256,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,768,0.007938133180141449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,512,0.007524266839027405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,128,0.002312533309062322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,128,0.014620799819628397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,256,0.0071168000499407455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,64,0.0022613334159056345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,128,0.006763733426729838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,512,32,0.0022645334402720133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,64,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,65536,0.011601066589355469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,512,32,0.01411626636981964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,16384,0.006762666503588359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,65536,0.02696640094121297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,16384,0.022309333086013794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,12288,0.00555626650651296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,12288,0.021630932887395225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,12288,0.02832213242848714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,10240,0.0059562668204307554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,16384,0.035028266906738284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,10240,0.01986773411432902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,8192,0.0059445331494013464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,8192,0.02212160031000773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,10240,0.02432960073153178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,7168,0.00544213354587555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,8192,0.02081386645634969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,7168,0.019246933857599895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,6144,0.00537066658337911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,6144,0.019058134158452353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,7168,0.019428267081578573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,6144,0.01691626707712809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,5120,0.005161599814891815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,5120,0.01867413322130839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,4096,0.00544213354587555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,4096,0.018454400698343913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,5120,0.015435733397801719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,4096,0.013245866696039835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,3584,0.005551999807357788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,3584,0.016539733608563742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,3072,0.005032533407211303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,3072,0.01681493322054545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,3584,0.01253973344961802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,2560,0.005436799923578898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,512,65536,0.12035733064015705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,3072,0.011606400211652119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,2560,0.017012266318003337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,2048,0.004753066599369049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,2048,0.014688000082969666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,1536,0.0038165333370367683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,1536,0.01527253290017446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,2560,0.011250133315722149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,2048,0.009961600104967754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,1024,0.0031317333380381264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,1024,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,768,0.0030432000756263735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,1536,0.009610666831334432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,768,0.014456533392270408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,512,0.0026133333643277483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,1024,0.007987200220425924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,768,0.007935999830563863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,512,0.015077333648999533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,256,0.002299733211596807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,256,0.014386133352915446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,512,0.007484800120194752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,128,0.00222080002228419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,128,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,256,0.0071263998746871945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,64,0.0022005334496498106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,128,0.0066538666685422255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,64,0.013457066814104714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,256,32,0.0021717332303524016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,65536,0.007974400122960409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,256,32,0.013673599561055502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,65536,0.02674986720085144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,16384,0.0059797331690788266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,16384,0.02239039937655131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,16384,0.03466666539510091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,12288,0.00553706685702006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,12288,0.020840533574422202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,10240,0.005881600081920624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,256,65536,0.11984533468882244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,10240,0.019800533850987755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,8192,0.005857066810131073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,8192,0.01874879995981852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,10240,0.024360533555348715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,12288,0.02768426736195882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,7168,0.005471999943256378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,8192,0.020985599358876547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,6144,0.005478399991989136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,7168,0.02028586665789286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,6144,0.01917653282483419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,5120,0.005098666747411092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,5120,0.01921280026435852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,7168,0.01936639944712321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,6144,0.01697173317273458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,4096,0.004670933385690053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,4096,0.018228266636530557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,5120,0.01530239979426066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,3584,0.005058133105436961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,3584,0.016777600844701132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,3072,0.005132799843947092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,4096,0.013707733154296875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,3072,0.016668800512949625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,2560,0.0050784001747767125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,3584,0.01249066690603892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,2560,0.01641386648019155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,2048,0.005044266581535339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,2048,0.016634666919708253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,3072,0.012036266922950744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,1536,0.0038058665891488397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,2560,0.01079360047976176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,1536,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,1024,0.003054933249950409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,2048,0.009979733824729919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,1024,0.014512000481287637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,1536,0.00921066701412201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,768,0.0030271999537944795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,1024,0.00830506682395935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,768,0.014051199952761332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,512,0.0026549334327379864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,768,0.007923200229803721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,512,0.015631999572118124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,256,0.0022229333718617756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,512,0.007096533477306366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,256,0.013848533233006796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,128,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,256,0.007524266839027405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,65536,0.11947626272837322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,64,0.002107733239730199
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,128,0.014270933469136557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,128,32,0.0021685334543387097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,64,128,128,0.007113599777221679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,65536,0.007516799867153168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,64,0.014127999544143677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,128,32,0.012583466370900473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,16384,0.005890133480230967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,65536,0.02669866681098938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,12288,0.005456000069777171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,10240,0.005131733417510986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,16384,0.020742400487263998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,12288,0.020870399475097657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,8192,0.005474133292833964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,7168,0.005103999873002371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,10240,0.01885120073954264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,6144,0.005070933202902476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,8192,0.018707199891408285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,7168,0.0196288009484609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,5120,0.00505920002857844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,4096,0.0046623999873797095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,6144,0.018793600797653198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,5120,0.01835839947064718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,3584,0.004990933338801066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,4096,0.017946666479110716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,3072,0.004664533336957296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,2560,0.005061333378156027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,3584,0.016482133666674295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,3072,0.016695467631022136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,2048,0.004666666686534882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,1536,0.003692800054947535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,2560,0.015774933497111003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,1024,0.003019733230272929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,2048,0.01560426652431488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,768,0.002593066543340683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,1536,0.015526400009791056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,512,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,1024,0.01482133368651072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,768,0.01443839967250824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,256,0.002229333420594533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,512,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,128,0.0022805333137512207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,64,0.0018496000518401462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,256,0.014190933108329773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,64,32,0.001794133335351944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,128,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,65536,0.009640533725420635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,64,0.014318933089574179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,64,32,0.012817066907882691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,16384,0.005940266450246175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,12288,0.005513600011666616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,65536,0.026763733228047686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,16384,0.021117866039276123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,10240,0.0051125332713127134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,8192,0.005512533088525137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,12288,0.020477867126464842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,10240,0.01884053349494934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,7168,0.005153066913286845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,6144,0.0050805335243542995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,8192,0.0189301331837972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,7168,0.018514132499694823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,5120,0.004633600016434988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,4096,0.004270933568477631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,6144,0.018114133675893148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,5120,0.018603734175364175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,3584,0.004624000191688538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,4096,0.0191103994846344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,3072,0.004286933441956838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,3584,0.016581333676973977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,2560,0.004673066735267639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,2048,0.004250666499137879
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,3072,0.016587733229001363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,2560,0.01548373301823934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,1536,0.0038602667550245917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,1024,0.003047466774781545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,2048,0.01637440025806427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,1536,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,768,0.002614400039116542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,512,0.002250666668017705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,1024,0.014512000481287637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,768,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,256,0.002288000037272771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,128,0.0022528000175952912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,512,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,256,0.014106667041778565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,64,0.0018570666511853536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,64,32,32,0.0017493333667516708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,128,0.014418133099873862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,64,0.013961600263913474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,64,32,32,0.012632532914479574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,16384,0.20385066668192545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,16384,0.36253439585367836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,12288,0.2681674639383952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,12288,0.1504543940226237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,10240,0.12796053091684978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,10240,0.2251551946004232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,8192,0.18729173342386882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,8192,0.11357440153757732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,8192,0.06453226804733277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,7168,0.15758293469746906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,7168,0.09602666695912679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,10240,0.07211093107859293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,6144,0.13608746528625487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,6144,0.08549119631449381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,12288,0.08882986704508464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,5120,0.1141856034596761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,5120,0.07490453720092774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,6144,0.04954986572265625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,4096,0.09264319737752279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,4096,0.06389333407084147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,7168,0.057127467791239415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,3584,0.08101759751637777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,3584,0.05669973293940226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,5120,0.04598613182703654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,3072,0.07119999726613363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,3072,0.0504095991452535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,16384,0.11606400012969971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,2560,0.059070932865142825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,2560,0.04585599899291992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,3584,0.03319573402404785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,2048,0.05338559945424398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,2048,0.0427562673886617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,4096,0.035523200035095216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,1536,0.037676799297332766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,1536,0.03384000062942505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,3072,0.030483200152715044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,1024,0.027485867341359455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,1024,0.026528000831604004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,2560,0.02653333346048991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,768,0.020708266894022623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,768,0.024270933866500855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,1536,0.017794134219487508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,512,0.01432319978872935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,512,0.02074346741040548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,2048,0.021911466121673585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,256,0.008350933591524761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,256,0.017372800906499227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,768,0.01241386632124583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,256,0.009565866986910502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,128,0.005836800237496694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,128,0.016557866334915163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,64,0.005100800096988678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,64,0.01653333306312561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,65536,32,0.005468800167242686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,65536,32,0.01850559910138448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,65536,0.3463775952657064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,65536,0.21029973030090332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,1024,0.01458133359750112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,16384,0.08738239606221518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,16384,0.05830826759338379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,128,0.008552533388137818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,12288,0.06142400105794271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,12288,0.048545066515604654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,65536,512,0.010478933652242024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,10240,0.05239253441492716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,10240,0.043194667498270674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,10240,0.0288480003674825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,8192,0.04326080083847046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,8192,0.037766400973002115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,12288,0.032927999893824264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,7168,0.037164799372355145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,7168,0.03346773386001587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,16384,0.040852268536885575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,6144,0.03279573321342468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,6144,0.0308896005153656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,8192,0.02409600019454956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,5120,0.02787733276685079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,5120,0.029009066025416058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,7168,0.02145493427912394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,4096,0.02294399936993917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,4096,0.024780799945195518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,5120,0.0173962672551473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,3584,0.020246400435765585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,3584,0.025704532861709595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,6144,0.01955519914627075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,3072,0.01762666702270508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,3072,0.022705066204071044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,4096,0.015197867155075073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,2560,0.015387733777364096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,2560,0.022362667322158813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,2560,0.012044800321261089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,2048,0.012309333682060242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,2048,0.020542933543523153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,3072,0.013962666193644205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,1536,0.009983999530474345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,1536,0.018481065829594932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,3584,0.014455466469128927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,1024,0.007613866527875264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,1024,0.01642346680164337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,1024,0.008460799853007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,768,0.0062837332487106325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,768,0.014651733636856078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,1536,0.00939520001411438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,512,0.003692800054947535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,512,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,2048,0.010552533467610677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,256,0.0030741333961486817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,256,0.0147189329067866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,256,0.007084799806276958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,128,0.0026709333062171934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,128,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,512,0.007922133306662242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,64,0.002593066543340683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,64,0.014398933450380961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,16384,32,0.002566399921973546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,16384,32,0.014432000120480857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,65536,0.25811519622802737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,128,0.006899199883143107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,65536,0.1598314603169759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,768,0.008090666433175405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,16384,0.07398719787597656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,16384,0.05279359817504883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,12288,0.05042879978815714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,12288,0.04278080066045125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,12288,0.030746666590372722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,10240,0.042871467272440594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,10240,0.03791786829630534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,16384,0.038675200939178464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,8192,0.035538132985432944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,8192,0.03248106638590495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,10240,0.02664746642112732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,7168,0.030538666248321533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,7168,0.03152640064557393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,8192,0.022284799814224245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,6144,0.02718186577161153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,6144,0.027517867088317872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,7168,0.02033920089403788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,5120,0.023063466946283976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,5120,0.02866986592610677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,6144,0.01826453407605489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,4096,0.019022933642069497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,4096,0.024794665972391765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,5120,0.017092265685399375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,3584,0.016926934321721397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,3584,0.023195733626683555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,4096,0.014594133694966635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,3072,0.014587733149528503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,3072,0.02225173314412435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,3584,0.013751467068990072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,2560,0.013317333658536276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,2560,0.020870399475097657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,3072,0.012477866808573405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,2048,0.011399466792742412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,2048,0.020348799228668214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,2560,0.011660800377527872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,1536,0.009612799684206644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,1536,0.018476800123850504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,1536,0.009257599711418152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,1024,0.007197866837183635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,1024,0.01636693378289541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,2048,0.010379733641942342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,768,0.005426133175690969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,768,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,1024,0.008463999629020691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,512,0.0042015999555587765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,512,0.014778666694959006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,768,0.007998933394749958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,256,0.0034304000437259674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,256,0.014444800217946372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,256,0.007089066505432129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,128,0.0029898665845394133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,128,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,512,0.007892266909281413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,64,0.002703999976317088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,64,0.014685866236686707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,12288,32,0.003020799905061722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,12288,32,0.014696533481280008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,128,0.007169066866238911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,65536,0.14618026415507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,65536,0.20973760286966958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,16384,0.06125119924545288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,16384,0.04415253400802612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,16384,0.03778666655222575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,12288,0.039892268180847165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,12288,0.03776959975560506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,12288,0.029787733157475786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,10240,0.03405439853668213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,10240,0.035368533929189046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,16384,65536,0.14833067258199056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,8192,0.028016000986099243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,8192,0.02983466585477193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,10240,0.025995733340581258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,7168,0.024570665756861367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,7168,0.029038933912913005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,8192,0.022132267554601036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,6144,0.021921066443125407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,6144,0.02508266568183899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,12288,65536,0.13248213132222492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,5120,0.018947199980417887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,5120,0.02342080076535543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,7168,0.020243199666341145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,4096,0.015230933825174967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,4096,0.023151999711990355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,6144,0.018658133347829182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,3584,0.013526399930318197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,3584,0.021923200289408366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,4096,0.014658133188883463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,3072,0.012412800391515096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,3072,0.020773333311080933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,5120,0.01657386620839437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,2560,0.01076586643854777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,2560,0.01900906761487325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,2560,0.010847999652226766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,2048,0.009175466497739156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,2048,0.018270933628082277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,3584,0.013692800203959146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,1536,0.007226666808128357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,1536,0.016607999801635742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,3072,0.012451199690500896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,1024,0.005145599941412607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,1024,0.015396266182263692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,1024,0.00844053328037262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,768,0.003912533322970072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,768,0.014971733093261719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,1536,0.00963200032711029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,512,0.0034122665723164878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,512,0.014779733618100485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,768,0.007950933277606964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,256,0.0030346666773160298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,256,0.014450132846832275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,512,0.007521066566308339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,128,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,128,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,2048,0.00995199978351593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,64,0.0025898667673269907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,256,0.007121066749095917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,64,0.014510933558146158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,10240,32,0.00264533335963885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,128,0.007112533350785573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,10240,32,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,65536,0.17859733899434407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,65536,0.11843307018280029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,16384,0.046632532278696695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,16384,0.04101226727167766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,12288,0.03516693512598674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,12288,0.033741867542266844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,12288,0.029314132531483968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,10240,0.030113067229588824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,10240,0.031454932689666745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,16384,0.0372927983601888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,8192,0.02512960036595662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,8192,0.02717439929644267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,10240,0.0252074658870697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,7168,0.021654399236043294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,7168,0.025154133637746174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,8192,0.021919999519983926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,6144,0.019466666380564372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,6144,0.024114133914311726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,7168,0.02005973259607951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,5120,0.01644159952799479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,5120,0.023312000433603923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,10240,65536,0.12764693101247152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,4096,0.013678933183352152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,4096,0.02178986668586731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,6144,0.018617600202560425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,3584,0.01214400033156077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,3584,0.021972266832987468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,5120,0.016598400473594666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,3072,0.010813867052396137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,3072,0.01953173279762268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,3584,0.013310933113098144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,2560,0.00946666697661082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,2560,0.018718934059143065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,4096,0.014122666915257773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,2048,0.007993599772453308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,2048,0.016669867436091106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,2048,0.010353066523869832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,1536,0.006698666512966156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,1536,0.017058134078979492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,2560,0.0113237331310908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,1024,0.004814933240413666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,1024,0.014999467134475707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,3072,0.012160000205039979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,768,0.003960533440113068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,768,0.014565333724021912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,768,0.007990399996439617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,512,0.0034933333595593774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,512,0.014532267053922018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,1024,0.008470400174458822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,256,0.00297173336148262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,256,0.014578133821487427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,256,0.0075647999842961625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,128,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,1536,0.009178666273752849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,128,0.01455466647942861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,64,0.0023872000475724537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,64,0.014586666226387024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,8192,32,0.0026464000344276427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,512,0.0076000000039736434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,8192,32,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,128,0.0067669332027435304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,65536,0.15410985946655273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,65536,0.10339306990305583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,16384,0.04447466532389323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,16384,0.04199680089950562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,12288,0.03416533470153808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,12288,0.032263465722401935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,12288,0.029260800282160444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,10240,0.029713066418965657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,10240,0.029150933027267456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,16384,0.03626453479131063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,8192,0.0233024001121521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,8192,0.026925865809122724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,10240,0.025602134068806966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,7168,0.019898666938145956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,7168,0.024623999993006386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,8192,0.021627734104792275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,6144,0.017998933792114258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,6144,0.023075199127197264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,7168,0.020273067553838096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,5120,0.01535146633783976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,5120,0.0228383998076121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,6144,0.018024533987045288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,4096,0.01283093293507894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,4096,0.02266133427619934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,5120,0.016158933440844216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,3584,0.011347200473149617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,3584,0.02150933345158895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,4096,0.013645866513252258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,3072,0.010414933164914448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,3072,0.019014400243759156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,3584,0.012899200121561686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,2560,0.00897813340028127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,2560,0.017072000106175742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,3072,0.012006400028864543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,2048,0.007844266792138416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,2048,0.0178272008895874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,2560,0.01121493379275004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,1536,0.006293333570162455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,1536,0.016193067034085594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,2048,0.010376532872517902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,1024,0.0042463997999827065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,1024,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,1536,0.009161600470542907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,768,0.0037823999921480812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,768,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,1024,0.008318933347860973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,512,0.003107200066248576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,512,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,768,0.00795839975277583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,256,0.0026517334083716077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,512,0.007515733440717061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,256,0.014762666821479798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,128,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,128,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,128,0.007102933526039123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,64,0.002233600119749705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,64,0.014326399564743042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,7168,32,0.002373333275318146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,7168,32,0.0141567995150884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,256,0.007108266651630402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,65536,0.12096426486968995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,65536,0.09666666984558106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,16384,0.03747306664784749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,16384,0.03715200026830037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,8192,65536,0.12645013332366944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,12288,0.028595199187596638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,12288,0.03238826592763265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,16384,0.036185598373413085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,10240,0.02513279914855957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,10240,0.027425066630045576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,12288,0.028853332996368407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,8192,0.021100799242655434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,8192,0.026500266790390015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,10240,0.025320533911387128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,7168,0.018796799580256145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,7168,0.023364265759785972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,8192,0.02170133392016093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,6144,0.017280000448226928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,6144,0.022716800371805825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,6144,0.018242132663726807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,5120,0.014854400356610616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,5120,0.022844799359639487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,7168,0.019836799303690592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,4096,0.012245333194732666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,4096,0.020126932859420778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,4096,0.013709867000579834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,3584,0.011027200023333232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,3584,0.01874133348464966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,5120,0.016123732924461363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,3072,0.009993599851926167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,3072,0.01763199965159098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,3584,0.012812800208727517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,2560,0.008760533730189006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,2560,0.016977065801620485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,3072,0.011762133240699768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,2048,0.00665280024210612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,2048,0.016787199179331462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,2560,0.010775466759999592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,1536,0.005047466854254405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,2048,0.010362666845321656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,1536,0.01577173372109731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,1024,0.0038453333079814913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,1024,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,7168,65536,0.12573546568552654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,768,0.003457066665093104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,768,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,1024,0.008506666620572407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,512,0.003078400095303853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,512,0.014542933305104574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,1536,0.009337600072224934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,768,0.007912533481915791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,256,0.002868266652027766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,256,0.013804800311724343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,128,0.002593066543340683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,128,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,512,0.007551999886830647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,64,0.0022944000860055285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,64,0.014241066575050355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,6144,32,0.0025194667279720305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,128,0.006716800232728322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,6144,32,0.014154666662216186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,65536,0.10562453269958497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,256,0.007179733117421467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,16384,0.03434240023295085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,65536,0.0886240005493164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,16384,0.03428479830423991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,12288,0.025967999299367266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,12288,0.029893332719802858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,12288,0.028588799635569255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,10240,0.021717333793640138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,10240,0.027483733495076497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,16384,0.03583893378575643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,8192,0.018225065867106118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,8192,0.024449066321055094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,10240,0.025183999538421632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,7168,0.01625386675198873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,7168,0.02335253357887268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,8192,0.021436800559361778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,6144,0.0146506667137146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,6144,0.02247999906539917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,7168,0.01986560026804606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,5120,0.013224533200263977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,5120,0.02123519976933797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,6144,0.017799466848373413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,4096,0.010993066430091857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,4096,0.01871253252029419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,5120,0.016139733791351318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,3584,0.010402133067448933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,3584,0.018513067563374837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,4096,0.013309866189956665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,3072,0.008866133292516072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,3072,0.016516266266504924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,3584,0.012505599856376648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,2560,0.007952000200748443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,2560,0.01693333387374878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,3072,0.012004266182581585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,2048,0.006717866659164429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,2048,0.015436800320943198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,2560,0.011245866616566975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,1536,0.004676266511281332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,1536,0.01627840002377828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,1536,0.009227733810742695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,1024,0.0038933334251244865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,1024,0.014697600404421488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,2048,0.009975467125574749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,768,0.0034101332227389016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,768,0.014447999993960061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,6144,65536,0.12608106931050617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,512,0.0030026666820049284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,512,0.014467199643452963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,768,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,256,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,256,0.014173866311709086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,1024,0.008398933211962382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,128,0.0026101333399613695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,128,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,128,0.007118933399518331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,64,0.0025568000972270967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,512,0.007492266595363617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,5120,32,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,64,0.014427733421325684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,256,0.007045333087444305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,5120,32,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,65536,0.08575572967529296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,65536,0.07415680090586343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,16384,0.027893332640329997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,16384,0.030398933092753093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,12288,0.02107093334197998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,12288,0.027412267525990804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,12288,0.02846933404604594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,10240,0.01837653319040934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,10240,0.023770666122436522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,16384,0.03578773339589437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,8192,0.015430399775505066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,8192,0.022817067305246987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,10240,0.025575466950734455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,7168,0.014057599504788718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,7168,0.02130026618639628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,8192,0.02121386726697286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,6144,0.012534399827321371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,6144,0.02102400064468384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,7168,0.019356799125671387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,5120,0.01032533347606659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,5120,0.01877546707789103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,6144,0.01738133430480957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,4096,0.00874133308728536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,4096,0.018496000766754152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,5120,0.015799466768900552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,3584,0.007952000200748443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,3584,0.017762132485707603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,4096,0.01372053325176239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,3072,0.007209600011507671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,3072,0.016948266824086507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,3584,0.012454400459925335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,2560,0.005913599828879039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,2560,0.016237866878509522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,3072,0.01172266701857249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,2048,0.005064533154169718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,2048,0.016503467162450155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,2560,0.010802132884661357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,1536,0.004043733328580856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,1536,0.015749333302179973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,1536,0.009181867043177288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,1024,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,1024,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,2048,0.010019200046857198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,768,0.0030613332986831666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,768,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,1024,0.007976533472537994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,512,0.0029333333174387617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,512,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,768,0.007965866724650066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,256,0.0026378666361172995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,256,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,256,0.007066666583220164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,128,0.0023061332603295645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,512,0.007513600091139476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,128,0.014556800325711569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,64,0.002403199921051661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,128,0.0067114666104316715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,64,0.014407466848691305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,4096,32,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,4096,32,0.014435199896494546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,65536,0.07443946997324626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,65536,0.06465706825256348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,16384,0.024515199661254882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,16384,0.03032960096995036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,5120,65536,0.12536319891611736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,12288,0.018201599518458046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,12288,0.02558186650276184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,12288,0.028808534145355225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,10240,0.01579839984575907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,10240,0.02287893295288086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,16384,0.036030932267506914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,8192,0.013796266913414002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,8192,0.02207466761271159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,10240,0.02517333428064982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,7168,0.013357866803805032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,7168,0.021545600891113282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,8192,0.021226666371027627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,6144,0.011756799618403117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,6144,0.020467199881871543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,6144,0.017463467518488564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,5120,0.009979733824729919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,5120,0.01875200072924296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,7168,0.01899306575457255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,4096,0.008354133367538452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,4096,0.017245866854985557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,4096,0.013286399841308593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,3584,0.007578666508197785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,3584,0.01653866668542226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,5120,0.015991466244061787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,3072,0.006949333349863689
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,3072,0.01811520059903463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,4096,65536,0.125709867477417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,2560,0.005898666878541311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,2560,0.016334933042526246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,3584,0.01283093293507894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,2048,0.004658133288224538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,2048,0.01611199975013733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,3072,0.012157866358757019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,1536,0.004508799811204275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,1536,0.016354133685429893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,1536,0.009286399682362874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,1024,0.003571200122435888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,1024,0.015083733201026916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,2048,0.009964799880981446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,768,0.0030271999537944795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,768,0.014490666985511779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,2560,0.011333333452542622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,512,0.003115733216206233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,512,0.014638933539390563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,512,0.007619200150171916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,256,0.00271573339899381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,768,0.007514666517575581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,256,0.014599466323852539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,128,0.0023029332359631857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,128,0.013893333077430726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,1024,0.008411733309427898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,64,0.0022762666145960493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,64,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3584,32,0.002293333411216736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,256,0.007118933399518331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,128,0.007146666447321574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3584,32,0.014499200383822122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,65536,0.06630293528238931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,65536,0.0618837316830953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,16384,0.021075199047724404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,16384,0.02802240053812663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,12288,0.016340266664822897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,12288,0.025172267357508344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,12288,0.02895039916038513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,10240,0.014602667093276978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,10240,0.023140267531077067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,16384,0.0358517328898112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,8192,0.012151466806729634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,8192,0.020859734217325846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,10240,0.02499413291613261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,7168,0.011518933375676473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,7168,0.021316266059875487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,8192,0.020750933885574342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,6144,0.010443733135859171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,6144,0.019517866770426433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,7168,0.019576533635457357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,5120,0.009567999839782714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,5120,0.019040000438690186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,6144,0.0173962672551473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,4096,0.00795839975277583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,4096,0.01774079998334249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,5120,0.01586133340994517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,3584,0.007379200061162312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,3584,0.016673066218694053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,4096,0.013739732901255288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,3072,0.0066431999206542965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,3072,0.01708586613337199
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,3584,0.013054933150609335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,2560,0.0059914668401082356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,2560,0.01665386656920115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,3072,0.012040533622105916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3584,65536,0.12554453214009603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,2048,0.004620799918969473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,2048,0.016723199685414632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,1536,0.004046933352947235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,1536,0.01493119994799296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,2560,0.010867200295130412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,1024,0.0034453332424163817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,1024,0.014668800433476768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,1536,0.009179733196894328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,768,0.002993066608905792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,1024,0.008397866288820903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,768,0.014513066411018372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,2048,0.010072533289591472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,512,0.002644266684850057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,512,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,256,0.002609066665172577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,256,0.014025599757830302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,256,0.0071168000499407455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,512,0.007500799993673961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,768,0.007956266899903615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,128,0.002219733347495397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,64,0.0021759999295075734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,128,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,3072,32,0.002165333429972331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,64,0.01400106648604075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,3072,32,0.013796266913414002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,128,0.00676800012588501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,65536,0.05971946716308594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,65536,0.05630079905192057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,16384,0.018236800034840902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,16384,0.025844266017278034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,12288,0.014865066607793173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,12288,0.024010666211446128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,12288,0.028472532828648884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,10240,0.013716266552607218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,10240,0.022738132874170938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,16384,0.0358240008354187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,8192,0.011363200346628825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,8192,0.020884267489115396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,10240,0.024363734324773154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,7168,0.012066133817036946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,7168,0.020516266425450645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,8192,0.020844799280166627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,6144,0.010245333115259807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,6144,0.018846933046976724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,7168,0.018981333573659262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,5120,0.008835200468699138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,5120,0.018642133474349974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,6144,0.017298134167989095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,4096,0.007786666850248973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,4096,0.017044266064961754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,5120,0.0153546671072642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,3584,0.00701333334048589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,3584,0.018101332585016887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,4096,0.01378986636797587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,3072,0.0064533332983652755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,3072,0.016727467377980552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,3584,0.012560000022252401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,2560,0.005457066496213277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,2560,0.015384533007939658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,3072,0.01172266701857249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,2048,0.004428799947102865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,2048,0.015803733468055726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,2560,0.010760533809661865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,1536,0.0038272000849246977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,1536,0.01798506577809652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,1536,0.009141332904497783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,1024,0.003398400048414866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,1024,0.014451199769973755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,2048,0.010070400436719258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,768,0.0031093334158261614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,768,0.014634666840235391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,3072,65536,0.12564266522725423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,512,0.002690133452415466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,1024,0.008317866424719492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,512,0.014654933412869772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,256,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,256,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,768,0.007924266656239827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,128,0.0022474666436513263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,128,0.014634666840235391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,256,0.007212799787521362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,512,0.007482666770617168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,64,0.0022613334159056345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2560,32,0.0022229333718617756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,128,0.006729599833488464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,64,0.0141184002161026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2560,32,0.014362667004267374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,65536,0.0457045316696167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,65536,0.051725868384043375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,16384,0.015307733416557312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,16384,0.025493333737055462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,12288,0.013683199882507324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,12288,0.022950400908788048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,12288,0.02765013376871745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,10240,0.012290133039156596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,10240,0.02284160057703654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,16384,0.03577280044555664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,8192,0.010943999886512757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,8192,0.02053546706835429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,10240,0.02442880074183146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,7168,0.01014293332894643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,7168,0.019192532698313395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,8192,0.020894932746887206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,6144,0.009581866860389709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,6144,0.01887786587079366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,7168,0.01904426614443461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,5120,0.008346666892369587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,5120,0.018667733669281004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,5120,0.015682133038838704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,4096,0.007113599777221679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,4096,0.01784106691678365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,6144,0.01767359972000122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,3584,0.006468266745408376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,3584,0.017493333419164023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,3584,0.012692266702651977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,3072,0.005558399856090546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,4096,0.013659733533859252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,3072,0.016708266735076905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,2560,0.005156266689300537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,2560,0.01585599978764852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,2560,0.010969600081443787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,2048,0.004740266501903534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,3072,0.011610666910807293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,2048,0.016566399733225504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,1536,0.004041599979003271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,1536,0.01567146678765615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2560,65536,0.1254325310389201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,1024,0.0036501333117485045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,1024,0.014659200112024942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,1536,0.008835200468699138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,768,0.0030378667016824085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,768,0.014551466703414917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,2048,0.01042133371035258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,1024,0.00842133363087972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,512,0.002678400029738744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,512,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,256,0.002609066665172577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,256,0.014574933052062988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,512,0.007095466554164887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,128,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,256,0.007243733108043671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,768,0.007965866724650066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,128,0.014502400159835815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,64,0.0022410665949185687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,2048,32,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,64,0.013718400398890176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,65536,0.03700586557388306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,2048,32,0.014179199934005737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,128,0.006795733173688252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,16384,0.013727999726931252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,65536,0.045767466227213546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,16384,0.024881066878636678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,12288,0.011293866237004598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,12288,0.02373866637547811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,12288,0.028124799331029255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,10240,0.010590933760007222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,10240,0.022746666272481283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,16384,0.03550933202107747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,8192,0.009272533655166625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,8192,0.02071040074030558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,10240,0.024940800666809083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,7168,0.008613333106040955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,7168,0.020526933670043945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,8192,0.02063573400179545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,6144,0.008746666709582011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,6144,0.018408532937367758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,7168,0.019578667481740315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,5120,0.008254933357238769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,5120,0.018071466684341432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,6144,0.016962132851282754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,4096,0.0070592001080513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,4096,0.01790293256441752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,4096,0.01325440009435018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,5120,0.01585706671079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,3584,0.0059349333246548975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,3584,0.016780799627304076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,3072,0.005558399856090546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,3072,0.01668693423271179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,3584,0.012538666526476542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,2560,0.005077333251635233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,3072,0.011848533153533935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,2560,0.01529813309510549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,2048,0.004453333218892416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,2048,0.016839466492335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,2048,0.009991466999053955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,1536,0.003802666564782461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,1536,0.015362133582433065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,2560,0.011326932907104492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,1024,0.0035616000493367515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,1024,0.014699733257293702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,2048,65536,0.12448000113169352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,768,0.0031445334355036415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,768,0.014722133676211039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,1536,0.008779733379681905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,1024,0.00844586690266927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,512,0.002603733291228612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,512,0.014669866363207499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,256,0.0026122666895389557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,256,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,768,0.008135466774304708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,512,0.007152000069618225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,128,0.002149333308140437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,128,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,256,0.007340799768765767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,128,0.006663466493288676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,64,0.002203733225663503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,64,0.013529599706331889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1536,32,0.002290133386850357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1536,32,0.013729066650072733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,65536,0.026552534103393553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,65536,0.03905066649119059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,16384,0.010390399893124899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,16384,0.023081600666046143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,12288,0.010492799679438274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,12288,0.021422932545344033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,12288,0.02868799964586894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,16384,0.03496640125910441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,10240,0.0095551997423172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,10240,0.02066346605618795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,8192,0.00839573343594869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,8192,0.01949653426806132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,8192,0.020641066630681357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,7168,0.007854933540026348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,7168,0.018617600202560425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,10240,0.024998400608698526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,6144,0.007375999788443248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,6144,0.019721599419911702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,7168,0.018649599949518838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,5120,0.006720000008742015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,5120,0.018722132841746012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,6144,0.017504000663757326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,4096,0.006424533327420552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,4096,0.016760534048080443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,5120,0.01530239979426066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,3584,0.005868799984455109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,3584,0.01669973333676656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,4096,0.013329066832860312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,3072,0.005648000041643778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,3072,0.017349332571029663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,3584,0.012839466333389282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,2560,0.0050453335046768185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,2560,0.015688533584276833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,3072,0.01209386686484019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,2048,0.004355200131734212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,2048,0.016692266861597697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,2560,0.010778666536013285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,1536,0.0038143999874591826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,1536,0.014769066373507181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,2048,0.010072533289591472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,1024,0.003475199888149897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,1024,0.01463573376337687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,1536,0.00876586635907491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,768,0.002977066735426585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,768,0.014444800217946372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,1024,0.00851200024286906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,512,0.002713600049416224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,512,0.014596266547838846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,768,0.00751146674156189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,256,0.0025727999707063037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,256,0.014151466886202493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,256,0.007096533477306366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,512,0.007572266459465027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,128,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,128,0.014412800470987955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,128,0.00631039987007777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,64,0.0022656001150608064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,1024,32,0.0022495999932289124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,64,0.0137855996688207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1536,65536,0.12411093711853027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,65536,0.023085866371790567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,1024,32,0.014497066537539164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,16384,0.010052266716957092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,65536,0.035097599029541016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,16384,0.02295253276824951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,12288,0.00877333382765452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,12288,0.021385600169499717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,12288,0.027847466866175334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,10240,0.008051200211048127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,10240,0.02138026754061381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,16384,0.03460373481114705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,8192,0.007227733234564463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,8192,0.020513067642847695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,10240,0.024437334140141806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,7168,0.00716480016708374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,7168,0.018655999501546224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,8192,0.020627200603485107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,6144,0.00653653343518575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,6144,0.0180949330329895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,7168,0.019028266270955406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,5120,0.006817066669464111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,5120,0.017667200167973837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,6144,0.017350399494171144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,4096,0.006685866912206013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,4096,0.01804800033569336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,5120,0.015639467040697734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,3584,0.005899733304977417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,1024,65536,0.12069013118743896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,3584,0.018942934274673463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,3072,0.0055071999629338585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,4096,0.013247999548912048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,3072,0.017083734273910522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,3584,0.012889599800109864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,2560,0.005076266825199127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,2560,0.015864533185958863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,3072,0.011618133385976155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,2048,0.004275199770927429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,2048,0.017292799552281697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,1536,0.00409706657131513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,1536,0.014854400356610616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,2560,0.010860799749692281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,1024,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,2048,0.009563733140627544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,1024,0.014589866995811463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,768,0.0030400000512599947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,768,0.015341867009798685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,1024,0.008326399823029835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,512,0.0025589334468046824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,768,0.00790293316046397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,512,0.014101333419481912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,1536,0.009215999643007915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,256,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,256,0.014124799768129984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,128,0.002251733342806498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,256,0.007123200098673503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,128,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,512,0.007462400197982788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,64,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,768,32,0.0022272000710169475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,64,0.014451199769973755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,128,0.006693333387374878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,65536,0.020037333170572914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,768,32,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,16384,0.008894933263460796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,65536,0.03271893262863159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,16384,0.020985599358876547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,12288,0.007133866846561432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,12288,0.02064639925956726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,12288,0.027691733837127686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,10240,0.006724266707897187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,10240,0.02160960038503011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,16384,0.03501439889272054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,8192,0.0059445331494013464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,8192,0.019511467218399046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,10240,0.02443199952443441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,7168,0.005873066683610281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,7168,0.01874666611353556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,8192,0.02058666745821635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,6144,0.005862399935722351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,6144,0.018889600038528444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,7168,0.018962132930755615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,5120,0.006704000135262807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,5120,0.016748799880345663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,6144,0.017391999562581383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,4096,0.005871999760468801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,4096,0.016757333278656007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,5120,0.015729066729545594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,3584,0.005859200159708659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,3584,0.016635732849438985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,4096,0.01362986663977305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,3072,0.005437866846720377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,3072,0.016810667514801026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,3584,0.012811733285586038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,2560,0.005017599960168203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,3072,0.012001066406567892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,2560,0.016434133052825928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,2048,0.004231466849644979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,2048,0.01576959987481435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,2560,0.010808533430099488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,1536,0.0037952000896135964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,2048,0.010011733571688334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,1536,0.016717867056528727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,1024,0.003052799900372823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,1024,0.014465066790580749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,768,65536,0.12027839819590251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,768,0.003126399964094162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,768,0.014552533626556396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,1536,0.00995306670665741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,512,0.00262719988822937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,1024,0.008384000261624653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,512,0.014591999848683677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,256,0.002333866556485494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,768,0.007524266839027405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,256,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,512,0.007160533467928569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,128,0.0023264000813166302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,128,0.014593066771825156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,256,0.006690133114655812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,64,0.0022261333962281544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,128,0.007130666573842366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,512,32,0.002198400100072225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,64,0.013612799843152366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,65536,0.012925866246223449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,512,32,0.014513066411018372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,16384,0.006340266764163971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,65536,0.02849493424097697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,16384,0.02074026664098104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,12288,0.006007466713587443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,12288,0.02058773239453634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,12288,0.028223999341328937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,16384,0.03459733327229818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,10240,0.005886933207511902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,10240,0.018846933046976724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,10240,0.02429973284403483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,8192,0.005482666691144307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,8192,0.01944213310877482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,7168,0.005936000247796377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,8192,0.02067626714706421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,7168,0.02064853310585022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,6144,0.006375466783841451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,6144,0.019024000565210978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,7168,0.018567466735839845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,5120,0.00581333339214325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,5120,0.01879146695137024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,6144,0.017480534315109254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,4096,0.005601066847642263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,4096,0.018475733200709023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,5120,0.01528320014476776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,3584,0.005118933320045471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,3584,0.0173962672551473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,512,65536,0.12037546634674072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,3072,0.005696000158786773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,3072,0.017069866259892784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,4096,0.014134400089581809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,2560,0.005234133203824361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,2560,0.01693120002746582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,3584,0.01284160017967224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,2048,0.004701866706212362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,2048,0.016406400005022685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,3072,0.011730133493741354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,1536,0.0039007999002933502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,2560,0.011252267162005107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,1536,0.01707306702931722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,1024,0.003110400090614955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,2048,0.009941333532333374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,1024,0.014773333072662353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,768,0.003052799900372823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,1536,0.00921066701412201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,1024,0.007918933530648549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,768,0.014654933412869772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,512,0.002643200010061264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,512,0.014619732896486918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,768,0.008004266520341237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,256,0.0022485333184401194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,512,0.007496533294518788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,128,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,256,0.014617600043614707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,128,0.014439466595649719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,256,0.007181866466999054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,64,0.002149333308140437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,256,32,0.0021461332837740582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,128,0.0071168000499407455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,64,0.01442026694615682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,65536,0.01123520036538442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,256,32,0.013923199971516928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,16384,0.005975466469923655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,65536,0.028823467095692952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,16384,0.020809600750605263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,12288,0.005432533224423727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,12288,0.019310933351516724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,256,65536,0.11976426442464191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,10240,0.005810133119424184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,10240,0.019078399737675986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,12288,0.02797973354657491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,8192,0.0054624001185099285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,16384,0.03469013373057048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,8192,0.018776534001032512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,7168,0.005109333495299021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,7168,0.01873706579208374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,10240,0.02435200015703837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,6144,0.005190399785836538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,6144,0.01842986742655436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,8192,0.020679465929667153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,5120,0.00508480022350947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,7168,0.018953599532445273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,5120,0.020478934049606323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,4096,0.005144533514976501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,4096,0.018776534001032512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,6144,0.017387733856836955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,3584,0.005208533505598704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,5120,0.015326933066050211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,3584,0.01735360026359558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,3072,0.004716800153255462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,4096,0.013707733154296875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,3584,0.012857600053151449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,3072,0.01625386675198873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,2560,0.004636799792448679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,2560,0.015726932883262636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,2048,0.0046528001626332605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,3072,0.011686399579048157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,2560,0.010773332913716634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,2048,0.014625066518783569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,1536,0.003457066665093104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,1536,0.017123200496037803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,1024,0.003011200080315272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,2048,0.009975467125574749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,1536,0.009134933352470398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,768,0.0026538667579491934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,1024,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,1024,0.008403199911117553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,768,0.014272000392278036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,512,0.002585600068171819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,768,0.007928533355395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,512,0.013703466455141703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,256,0.0022133332987626392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,256,0.012713600198427835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,512,0.007158400118350982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,128,0.0022240000466505687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,256,0.007095466554164887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,128,0.014410666624704995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,128,0.006679466863473256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,64,0.0019146667172511417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,64,0.012613333264986672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,128,32,0.002113066613674164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,128,32,0.01262933313846588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,65536,0.007993599772453308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,16384,0.005902933577696482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,65536,0.026155734062194826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,12288,0.0050240000089009605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,48,128,65536,0.11996160348256428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,16384,0.02104640007019043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,12288,0.018875734011332194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,10240,0.0060245335102081295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,8192,0.00544213354587555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,7168,0.005095466474692027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,10240,0.018770132462183634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,6144,0.005109333495299021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,8192,0.01883946657180786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,7168,0.018310399850209554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,5120,0.00526506652434667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,6144,0.017246933778127034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,5120,0.018895999590555827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,4096,0.004752000172932943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,3584,0.0046410664916038515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,3072,0.004273066421349844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,4096,0.0168938676516215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,3584,0.01809813380241394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,2560,0.004670933385690053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,3072,0.01606613298257192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,2048,0.004271999994913737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,1536,0.003487999985615412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,2560,0.01634239951769511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,2048,0.014732799927393594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,1024,0.003101866692304611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,1536,0.01535040040810903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,768,0.0031040000418821974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,1024,0.01442026694615682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,512,0.002739199995994568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,256,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,512,0.014203733205795288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,768,0.014130133390426635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,128,0.0021503999829292296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,64,0.0017535999417304994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,256,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,128,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,64,32,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,64,0.012733866771062216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,65536,0.007970133423805236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,64,32,0.012577066818873087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,16384,0.005930666625499725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,65536,0.026315732796986895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,12288,0.005142400165398916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,10240,0.00547733356555303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,16384,0.019562667608261107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,12288,0.01887893279393514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,8192,0.005470933516820272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,7168,0.005061333378156027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,10240,0.01875093380610148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,8192,0.017668267091115318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,6144,0.005206400156021118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,5120,0.005074133475621542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,7168,0.017443199952443443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,6144,0.01862186590830485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,4096,0.004227200150489807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,3584,0.004653866589069367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,5120,0.01838399966557821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,4096,0.01984213391939799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,2560,0.004305066665013632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,3072,0.004985600213209788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,3584,0.017755732933680216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,3072,0.016875733931859337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,2048,0.004816000163555145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,1536,0.003505066782236099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,2560,0.016301866372426352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,2048,0.014656000336011252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,1024,0.003017599880695343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,768,0.0026367999613285064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,512,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,1024,0.014460800091425577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,1536,0.01557866632938385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,768,0.01499626636505127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,256,0.002199466774861018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,128,0.002181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,512,0.013237333297729493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,64,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,256,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,48,32,32,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,128,0.014335999886194864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,64,0.0135861337184906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,48,32,32,0.01439573367436727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,16384,0.2025376001993815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,16384,0.3769205411275228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,12288,0.2673738797505697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,12288,0.1551424026489258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,10240,0.22814079920450844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,10240,0.12698453267415363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,8192,0.18553172747294108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,8192,0.11129066944122315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,8192,0.06279253164927165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,7168,0.1549887975056966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,7168,0.09321386814117431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,10240,0.07587733268737792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,6144,0.14202240308125813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,6144,0.08385492960611979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,12288,0.09561599890391032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,5120,0.11352852980295818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,5120,0.07022186915079752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,6144,0.04858133395512899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,4096,0.10012479623158772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,4096,0.06223146518071493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,7168,0.05727680126825968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,3584,0.07941760222117106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,3584,0.05432533423105875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,5120,0.042949334780375166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,3072,0.06925333340962728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,3072,0.049429333209991454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,16384,0.12477440039316814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,2560,0.058057598272959386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,2560,0.04331839879353841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,3584,0.032705066601435344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,2048,0.04761173327763875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,2048,0.03966079950332642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,4096,0.036962131659189865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,1536,0.03668053150177002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,1536,0.032731733719507855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,3072,0.03072106639544169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,1024,0.025630933046340943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,1024,0.02537279923756917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,2048,0.021950932343800862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,768,0.0196341335773468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,768,0.022817067305246987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,2560,0.02510720094045003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,512,0.013298133015632629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,512,0.019687465826670327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,1536,0.019026132424672444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,256,0.00720000018676122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,256,0.016607999801635742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,512,0.011194666226704914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,128,0.005064533154169718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,128,0.0164682666460673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,256,0.009574400385220845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,64,0.004275199770927429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,64,0.016641066471735636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,65536,32,0.004549333453178405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,65536,32,0.0182805339495341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,65536,0.3772351900736491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,65536,0.20199359258015953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,128,0.008752000331878663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,16384,0.08276693026224771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,16384,0.05549439986546835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,768,0.013159466783205667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,12288,0.059044265747070314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,12288,0.04729386568069458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,65536,1024,0.014869333306948344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,10240,0.04999573230743408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,10240,0.04200106859207153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,10240,0.029044266541798907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,8192,0.0409440000851949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,8192,0.0364192008972168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,12288,0.032682667175928756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,7168,0.036139734586079914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,7168,0.033404799302419026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,16384,0.04207573334376018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,6144,0.031751465797424314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,6144,0.03164586623509725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,8192,0.023948800563812257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,5120,0.02717439929644267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,5120,0.027929600079854327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,7168,0.022796799739201866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,4096,0.021919999519983926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,4096,0.02555626630783081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,5120,0.017364267508188883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,3584,0.01957013408342997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,3584,0.022937599817911783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,6144,0.021117866039276123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,3072,0.01699413259824117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,3072,0.02255679965019226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,4096,0.015150933464368185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,2560,0.014628266294797262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,2560,0.02244480053583781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,2560,0.012183466553688049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,2048,0.012477866808573405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,2048,0.020383999745051066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,3584,0.014988799889882406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,1536,0.009564800063769023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,1536,0.018589866161346436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,3072,0.014146133263905843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,1024,0.0075914666056633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,1024,0.017767467101415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,1024,0.008452266454696655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,1536,0.00960106650988261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,768,0.00589333325624466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,768,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,512,0.004216533402601878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,512,0.014867200454076131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,2048,0.011313066879908244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,256,0.0031040000418821974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,256,0.014727466305096946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,512,0.007627733548482259
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,128,0.0026591998835404714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,128,0.014556800325711569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,768,0.007948799928029378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,64,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,128,0.007108266651630402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,64,0.01448853313922882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,16384,32,0.002639999985694885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,16384,32,0.014451199769973755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,256,0.007502933343251546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,65536,0.24730134010314941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,65536,0.15043306350708008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,16384,0.06838933626810709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,16384,0.05337066650390625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,12288,0.05117546717325846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,12288,0.04535786708196004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,12288,0.030271999041239422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,10240,0.04422826766967773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,10240,0.040047999223073324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,16384,0.038369067509969074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,8192,0.03735893170038859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,8192,0.030833067496617635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,10240,0.026390399535497027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,7168,0.033037867148717245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,7168,0.028681600093841554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,8192,0.022383999824523926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,6144,0.028939733902613323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,6144,0.026951466004053754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,7168,0.020307199160257975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,5120,0.021917865673700968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,5120,0.026473599672317504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,6144,0.018599466482798258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,4096,0.01739413340886434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,4096,0.024474666515986124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,5120,0.016993065675099693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,3584,0.015718400478363037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,3584,0.022631466388702393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,4096,0.014637866616249084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,3072,0.013890133301417033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,3072,0.020641066630681357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,3584,0.013684266805648803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,2560,0.013049599528312684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,2560,0.020482132832209267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,3072,0.012521599729855856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,2048,0.010840533177057902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,2048,0.01887680093447367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,2560,0.011265066266059876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,1536,0.008763733506202697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,1536,0.016540799538294473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,1536,0.009195733070373534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,1024,0.006717866659164429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,1024,0.014632532993952433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,2048,0.010420266787211101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,768,0.005269333223501841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,768,0.014709333578745524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,1024,0.008361599842707316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,512,0.0037685332198937735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,512,0.01503679951032003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,768,0.008019199967384339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,256,0.0031167998909950255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,256,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,512,0.007516799867153168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,128,0.0026357332865397137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,128,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,256,0.007347199817498524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,64,0.002696533252795537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,64,0.01458453337351481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,12288,32,0.002674133330583572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,12288,32,0.01448853313922882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,128,0.006794666747252147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,65536,0.21955626805623374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,65536,0.14208213488260907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,16384,0.06000853379567465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,16384,0.05050133466720581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,16384,0.03744320074717204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,12288,0.044546135266621906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,12288,0.040394667784372965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,12288,0.02932586669921875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,10240,0.03829546769460042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,10240,0.034482133388519284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,16384,65536,0.14839253425598145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,8192,0.03233493367830913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,8192,0.03084160089492798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,10240,0.025569067398707075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,7168,0.025628799200057985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,7168,0.02704533338546753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,12288,65536,0.1324287970860799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,6144,0.02249600092569987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,6144,0.026624000072479247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,8192,0.0219050665696462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,5120,0.019062399864196777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,5120,0.023851732412974037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,7168,0.020201599597930907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,4096,0.016165332992871602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,4096,0.023865600426991783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,4096,0.014181333780288696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,3584,0.014215466380119324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,3584,0.02090346614519755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,5120,0.017192532618840538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,3072,0.012617599964141846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,3072,0.02074986696243286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,6144,0.018653867642084758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,2560,0.010959999759991963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,2560,0.021682133277257286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,2560,0.01127893328666687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,2048,0.010044800241788228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,2048,0.019814399878184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,3072,0.012472533186276754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,1536,0.008387200037638346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,1536,0.016620799899101257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,3584,0.013610666990280152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,1024,0.005940266450246175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,1024,0.016593066851298015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,1024,0.008330666522185007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,768,0.004781866570313772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,768,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,1536,0.009633066256841023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,512,0.003982933362325033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,512,0.014642133315404256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,768,0.007926400005817413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,256,0.003020799905061722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,256,0.014702933033307395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,512,0.007663999994595845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,128,0.002755200117826462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,256,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,128,0.014605866869290671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,64,0.00264533335963885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,2048,0.010456533233324686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,64,0.014539733529090881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,10240,32,0.0026506667335828146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,10240,32,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,65536,0.15998506546020508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,65536,0.10879786809285481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,16384,0.05149866739908854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,128,0.006650666892528534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,16384,0.03701653480529785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,12288,0.03115306695302327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,12288,0.03196266690889994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,12288,0.028940800825754804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,10240,0.026792534192403156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,10240,0.02881386677424113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,16384,0.037058134873708085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,8192,0.022323199113210044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,8192,0.02525866627693176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,10240,0.02569813330968221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,7168,0.01986133257548014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,7168,0.02327359914779663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,8192,0.021572266022364298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,6144,0.017400532960891724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,6144,0.023299199342727662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,7168,0.0202890674273173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,5120,0.014919466773668923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,5120,0.02224853237469991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,6144,0.018193066120147705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,4096,0.012088533242543538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,4096,0.019792000452677407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,10240,65536,0.1280234654744466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,3584,0.011256532867749532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,3584,0.019578667481740315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,5120,0.01656000018119812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,3072,0.010076799988746643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,3072,0.020475733280181884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,3584,0.013245866696039835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,2560,0.008821333448092144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,2560,0.0195413331190745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,4096,0.013765333096186319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,2048,0.007525333265463512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,2048,0.0162282665570577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,3072,0.012145066261291504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,1536,0.006761600077152252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,1536,0.01477226714293162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,2048,0.009983999530474345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,1024,0.004642133414745331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,2560,0.01090666651725769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,1024,0.016130133469899496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,768,0.0035434665779272715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,768,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,1536,0.009302399555842082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,512,0.0031839999059836066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,768,0.007499733567237854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,512,0.01467626690864563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,256,0.00275093341867129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,256,0.014725333452224732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,1024,0.008585600058237712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,128,0.002277333289384842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,256,0.007110400001207988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,128,0.014536533753077188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,64,0.002253866692384084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,64,0.013742933670679728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,8192,32,0.0023093332846959433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,8192,32,0.01381119986375173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,128,0.00710399995247523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,65536,0.15098133087158203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,512,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,65536,0.0969055970509847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,16384,0.042855465412139894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,16384,0.03800426721572876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,12288,0.033096534013748166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,12288,0.029284266630808513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,12288,0.028875732421875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,10240,0.027102933327356978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,10240,0.026895999908447266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,16384,0.03620479901631673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,8192,0.022399999698003135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,8192,0.025198932488759356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,10240,0.02541653315226237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,7168,0.019803732633590698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,7168,0.023465599616368613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,8192,0.021614933013916017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,6144,0.015938133001327515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,6144,0.022654932737350465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,7168,0.020089600483576456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,5120,0.013642666737238565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,5120,0.020618667205174766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,6144,0.018307199080785118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,4096,0.011275733510653179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,4096,0.018634666999181114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,5120,0.016214399536450704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,3584,0.010221866766611735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,3584,0.01991573373476664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,4096,0.013689600427945457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,3072,0.009169066945711773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,3072,0.020565332969029744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,3584,0.013172266880671182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,2560,0.007956266899903615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,2560,0.01825493375460307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,3072,0.011598933736483257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,2048,0.006980266670385997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,2048,0.015447466572125753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,2560,0.01084266702334086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,1536,0.006052266558011373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,1536,0.016846932967503867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,2048,0.00995306670665741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,1024,0.003874133278926214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,1024,0.016619732975959776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,1536,0.009191466371218364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,768,0.0034080001215140024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,768,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,768,0.007528533538182576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,512,0.0030271999537944795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,512,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,1024,0.007978666822115581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,256,0.0025909334421157838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,256,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,256,0.0070602665344874065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,128,0.002179199953873952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,128,0.014424533645311991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,512,0.007478400071461995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,64,0.0021920000513394673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,64,0.013910399874051413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,128,0.006764799853165944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,7168,32,0.0022122666239738466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,7168,32,0.01372160017490387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,65536,0.11605439980824787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,65536,0.09069546858469645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,16384,0.03554240067799886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,16384,0.03570346832275391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,8192,65536,0.12608959674835205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,12288,0.02730133334795634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,12288,0.030610134204228718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,16384,0.03581546545028687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,10240,0.024037333329518636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,12288,0.028838400046030683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,10240,0.027265065908432008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,8192,0.020453333854675293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,8192,0.024998400608698526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,8192,0.02156053384145101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,7168,0.018032000462214152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,7168,0.021502933899561563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,10240,0.025561600923538208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,6144,0.016085333625475564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,6144,0.021869866053263347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,7168,0.019797333081563315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,5120,0.014145066340764364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,5120,0.020399999618530274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,6144,0.018062933286031087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,4096,0.010461866855621338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,4096,0.01946880022684733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,5120,0.016126933693885803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,3584,0.009473066528638203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,3584,0.01866453289985657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,4096,0.013741866747538248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,3072,0.008569600184758504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,3072,0.018862932920455933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,3072,0.012025599678357441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,2560,0.007742933432261149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,2560,0.01796906590461731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,3584,0.012862933675448099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,2048,0.0066549330949783325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,2048,0.015100799997647605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,2048,0.009595732887585957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,1536,0.005569066603978475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,2560,0.0108842670917511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,1536,0.016459733247756958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,1024,0.003960533440113068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,1024,0.014641066392262777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,7168,65536,0.1256383975346883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,768,0.0033962666988372804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,768,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,1536,0.009324799974759419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,512,0.003107200066248576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,768,0.007918933530648549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,512,0.014731733004252115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,256,0.002696533252795537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,256,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,1024,0.008393599589665731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,128,0.002284800012906392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,128,0.0137855996688207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,128,0.006757333377997081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,64,0.002271999915440877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,256,0.007181866466999054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,64,0.013897599776585898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,6144,32,0.002338133255640666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,6144,32,0.013385599851608277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,65536,0.10445226828257244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,65536,0.08210453192392984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,16384,0.033088000615437825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,512,0.007107200225194295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,16384,0.03432533343633016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,12288,0.024769065777460735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,12288,0.029685332377751665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,12288,0.028542933861414592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,10240,0.021421867609024047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,10240,0.02698880036671956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,16384,0.035795199871063235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,8192,0.0184170663356781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,8192,0.02288853327433268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,10240,0.025172267357508344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,7168,0.016334933042526246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,7168,0.02271359960238139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,8192,0.02146773338317871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,6144,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,6144,0.02069013317426046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,7168,0.019959467649459838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,5120,0.012857600053151449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,5120,0.020473599433898926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,6144,0.017414400974909462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,4096,0.009562666217486065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,4096,0.018438400824864705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,5120,0.015887999534606935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,3584,0.00886400043964386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,3584,0.017939200003941856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,4096,0.013685333728790283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,3072,0.00793280005455017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,3072,0.019100799163182577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,3584,0.013081600268681845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,2560,0.007291733225186665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,2560,0.01665386656920115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,3072,0.011658666531244914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,2048,0.005861333509286245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,2048,0.01605226695537567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,2560,0.010877866546312969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,1536,0.005342933535575867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,1536,0.017089066902796428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,2048,0.00998293360074361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,1024,0.004212266703446706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,1024,0.014998400211334228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,1024,0.008337066570917765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,768,0.003621333340803782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,768,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,1536,0.00937493344148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,512,0.0031114667654037476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,512,0.014548266927401224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,6144,65536,0.12580160299936932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,256,0.0027274665733178455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,256,0.014591999848683677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,256,0.007546666761239369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,128,0.00264533335963885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,512,0.007602133353551229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,128,0.013963733116785684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,64,0.0022645334402720133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,768,0.007579733431339264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,64,0.014533332983652749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,5120,32,0.002145066608985265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,5120,32,0.014510933558146158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,65536,0.08700053691864014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,128,0.006715733309586842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,16384,0.0271722674369812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,65536,0.06739520231882731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,16384,0.0290175994237264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,12288,0.01946880022684733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,12288,0.026228266954421996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,12288,0.028868265946706134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,10240,0.017068799336751303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,10240,0.024740266799926757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,16384,0.035813331604003906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,8192,0.014113066593805948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,8192,0.022754132747650146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,10240,0.02518293261528015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,7168,0.01295146644115448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,7168,0.021185066302617392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,8192,0.0214901328086853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,6144,0.011985066533088683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,6144,0.02094506621360779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,7168,0.019375999768575035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,5120,0.011310933033625285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,5120,0.019774933656056724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,6144,0.01743146578470866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,4096,0.010079999764760334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,4096,0.019684267044067384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,5120,0.015705600380897522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,3584,0.00885973374048869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,3584,0.016544000307718913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,4096,0.013324800133705138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,3072,0.008277333279450735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,3072,0.01655359963575999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,3584,0.01288640002409617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,2560,0.007036800185839335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,2560,0.015547733505566916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,2560,0.01083626647790273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,2048,0.005127466718355815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,2048,0.01532906691233317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,3072,0.012018133203188579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,1536,0.003832533210515976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,1536,0.014692266782124838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,1536,0.00920853316783905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,1024,0.0034314667185147605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,1024,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,2048,0.010391466816266378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,768,0.0030048000315825146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,768,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,1024,0.008371200164159138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,512,0.002631466587384542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,512,0.01442346672217051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,768,0.007911466558774312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,256,0.0026079999903837843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,256,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,256,0.007111466427644093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,128,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,512,0.00755626658598582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,128,0.01409173309803009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,64,0.002143999934196472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,64,0.013781332969665527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,4096,32,0.0021727999051411946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,4096,32,0.013690666357676188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,128,0.006695466736952464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,65536,0.0754869302113851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,65536,0.060730667908986416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,16384,0.023439999421437582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,16384,0.02769920031229655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,5120,65536,0.12554986476898194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,12288,0.01781866749127706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,12288,0.02525866627693176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,12288,0.028869332869847615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,10240,0.015427199999491372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,10240,0.025062400102615356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,16384,0.035884801546732584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,8192,0.013313066959381104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,8192,0.02344640096028646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,10240,0.025102933247884113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,7168,0.012405332922935487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,7168,0.02108586629231771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,8192,0.021066667636235555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,6144,0.011291733384132386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,6144,0.020853332678476968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,7168,0.019033600886662803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,5120,0.011493333180745443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,5120,0.019793067375818887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,6144,0.017347200711568197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,4096,0.009803733229637146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,4096,0.018796799580256145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,4096,0.01368106702963511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,3584,0.008805333574612936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,3584,0.016563199957211814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,5120,0.015795200069745382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,3072,0.008080000181992848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,3072,0.0165802667538325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,3584,0.012923733393351237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,2560,0.006942933301130931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,2560,0.01683626572291056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,3072,0.011982933680216471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,2048,0.004852266609668731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,2048,0.014954666296641031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,4096,65536,0.12516586780548095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,1536,0.0038677332301934562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,1536,0.014512000481287637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,2048,0.010132267077763876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,1024,0.0034847999612490333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,1024,0.01535040040810903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,2560,0.010870400071144103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,768,0.0031338666876157125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,768,0.014595199624697366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,1536,0.009144533673922222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,512,0.002631466587384542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,512,0.014593066771825156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,768,0.008096000055472057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,256,0.002661333233118057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,512,0.007522133489449819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,256,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,1024,0.008434133728345235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,128,0.002299733211596807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,128,0.014587733149528503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,128,0.007197866837183635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,256,0.007102933526039123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,64,0.00222080002228419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3584,32,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,64,0.01366933286190033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,65536,0.06368853251139323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3584,32,0.013520000378290811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,65536,0.05670826832453409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,16384,0.01999573310216268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,16384,0.02722240090370178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,12288,0.01585919956366221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,12288,0.02637973427772522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,12288,0.02895466685295105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,10240,0.014353066682815552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,10240,0.023477333784103393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,16384,0.03586346705754598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,8192,0.012088533242543538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,8192,0.02103253404299418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,10240,0.024869332710901894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,7168,0.011345066626866658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,7168,0.021155200401941934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,8192,0.0206986665725708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,6144,0.010398933291435241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,6144,0.021384533246358237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,7168,0.019192532698313395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,5120,0.00965226689974467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,5120,0.0188810666402181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,6144,0.017358932892481485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,4096,0.008769067128499348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,4096,0.01651946703592936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,4096,0.013427199920018515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,3584,0.00820906658967336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,3584,0.016635732849438985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,5120,0.015878400206565856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,3072,0.007221333185831706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,3072,0.0172160009543101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,3584,0.012492799758911132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,2560,0.006037333110968272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,2560,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,3072,0.011755733688672384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,2048,0.004785066843032837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,2048,0.017406932512919106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3584,65536,0.12525973320007325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,1536,0.003869866579771042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,1536,0.014570666352907815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,2560,0.011148800452550251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,1024,0.0033632000287373864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,1024,0.014668800433476768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,2048,0.010194133718808491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,768,0.0031040000418821974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,768,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,1024,0.008257066706816356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,512,0.0026591998835404714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,768,0.007957333326339721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,512,0.014651733636856078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,256,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,256,0.014575999975204468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,1536,0.009225599964459737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,128,0.0022357332209746042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,128,0.014114133516947427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,256,0.007073066631952922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,128,0.006811733543872833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,64,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,3072,32,0.002163200080394745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,64,0.01367573340733846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,3072,32,0.01383680005868276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,512,0.007503999769687653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,65536,0.05519359906514486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,65536,0.052390400568644205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,16384,0.01861013372739156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,16384,0.025787732998530072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,12288,0.013623467087745667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,12288,0.023935999472935995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,12288,0.028135466575622558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,10240,0.012682666381200155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,10240,0.022754132747650146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,16384,0.03585066795349121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,8192,0.010998400052388509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,8192,0.022087466716766358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,10240,0.024782933791478477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,7168,0.010474666953086853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,7168,0.019994666179021202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,8192,0.02078826626141866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,6144,0.009925333658854167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,6144,0.01962453325589498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,7168,0.019025067488352455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,5120,0.008731733759244282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,5120,0.01797119975090027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,6144,0.017547732591629027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,4096,0.008187733093897502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,4096,0.016790399948755898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,5120,0.015723733107248943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,3584,0.007131733496983846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,3584,0.016618667046229045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,4096,0.013808000087738036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,3072,0.00661653329928716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,3072,0.01831679940223694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,3584,0.012840533256530761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,2560,0.005926399926344554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,2560,0.016254933675130208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,3072,0.011653332908948263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,2048,0.004293333490689596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,2048,0.01472106675306956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,2560,0.011178666353225708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,1536,0.003884800026814143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,1536,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,2048,0.010133333007494609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,1024,0.0034688000877698263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,1024,0.015539200107256571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,1536,0.00918506681919098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,768,0.0029877332349618276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,768,0.014444800217946372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,1024,0.00844373305638631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,512,0.002678400029738744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,512,0.014577066898345948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,3072,65536,0.125710932413737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,256,0.0025642665723959604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,768,0.007589333256085714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,256,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,128,0.002251733342806498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,512,0.007473066449165344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,128,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,64,0.0022240000466505687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,64,0.012910933295885722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,256,0.007084799806276958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2560,32,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2560,32,0.013106133540471396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,128,0.0070698668559392285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,65536,0.04499306678771973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,16384,0.015290666619936624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,65536,0.04673386812210083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,16384,0.025809067487716674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,12288,0.011852799852689107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,12288,0.02384106715520223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,12288,0.02821226716041565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,10240,0.010886399944623312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,10240,0.023507199684778848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,16384,0.03509333531061808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,8192,0.011246933539708456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,8192,0.020732800165812172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,10240,0.024979199965794882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,7168,0.01050986647605896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,7168,0.022461867332458495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,8192,0.020773333311080933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,6144,0.009562666217486065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,6144,0.019412267208099365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,7168,0.01921066641807556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,5120,0.008958933750788371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,5120,0.01773759921391805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,6144,0.016961065928141277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,4096,0.007794133325417836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,4096,0.016832000017166136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,5120,0.015838932991027833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,3584,0.0072543998559316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,4096,0.013244799772898355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,3584,0.0173962672551473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,3072,0.005752533177534739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,3072,0.015703466534614564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,3584,0.012800000111262002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,2560,0.005077333251635233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,3072,0.012134400010108948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,2560,0.015225600202878317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,2048,0.00425493319829305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,2048,0.014670933286348978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2560,65536,0.12565759817759198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,1536,0.004023466755946477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,2048,0.009971200426419576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,1536,0.014696533481280008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,1024,0.0033461332321166994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,2560,0.010930132865905762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,1024,0.015579733252525329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,768,0.00306986669699351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,768,0.014747732877731323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,768,0.007947733501593272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,1024,0.00840106705824534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,512,0.0027221334477265675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,512,0.01467626690864563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,256,0.002550400048494339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,1536,0.009180800120035807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,256,0.014762666821479798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,128,0.0022154666483402254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,128,0.014537599682807923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,256,0.0070783997575442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,64,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,512,0.007550933460394542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,2048,32,0.0022101332743962605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,64,0.013117866714795432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,128,0.007134933272997539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,65536,0.03561493158340454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,2048,32,0.013375999530156455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,65536,0.04100373188654582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,16384,0.014947199821472168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,16384,0.026169600089391072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,12288,0.012235732873280843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,12288,0.023350399732589722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,12288,0.02815786600112915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,10240,0.010971732934316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,10240,0.021802665789922078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,16384,0.03522666692733765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,8192,0.010001066327095031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,8192,0.01902079979578654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,10240,0.02476373314857483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,7168,0.009283199906349182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,7168,0.02100373307863871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,8192,0.02062079906463623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,6144,0.008460799853007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,6144,0.018637865781784058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,7168,0.01908586621284485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,5120,0.008008533219496409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,5120,0.018101332585016887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,6144,0.017340799172719322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,4096,0.006691200037797292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,5120,0.01535146633783976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,4096,0.01726933320363363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,3584,0.006295466423034668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,3584,0.017612799008687337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,3584,0.012869333227475485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,3072,0.005426133175690969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,3072,0.017799466848373413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,4096,0.01373546620210012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,2560,0.0050784001747767125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,2560,0.016809600591659545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,3072,0.011684266726175944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,2048,0.004851200183232625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,2048,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,2560,0.01086186667283376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,1536,0.0038762666285037995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,1536,0.01458133359750112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,1536,0.00920960009098053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,1024,0.003409066547950109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,2048,65536,0.12511680126190186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,1024,0.01511573294798533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,768,0.002997333308060964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,768,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,2048,0.00998293360074361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,512,0.0026421333352724713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,512,0.015084800124168397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,512,0.007080533107121785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,768,0.007963733375072479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,256,0.0024896000822385153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,1024,0.008344533046086629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,256,0.01460693379243215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,128,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,128,0.014380799730618796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,64,0.0021269333859284716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,64,0.014437333742777506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,128,0.007041066884994507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,256,0.007150933146476746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1536,32,0.0021354667842388153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,65536,0.025114667415618897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1536,32,0.012956800063451132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,16384,0.011657599608103435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,65536,0.03360746701558431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,16384,0.02179093360900879
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,12288,0.00977066655953725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,12288,0.020759467283884683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,12288,0.02815893292427063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,10240,0.00877333382765452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,10240,0.019387733936309815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,16384,0.035019731521606444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,8192,0.007869866490364075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,8192,0.019051732619603474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,10240,0.024714666604995727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,7168,0.0072736000021298725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,7168,0.019358932971954346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,8192,0.020733867088953653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,6144,0.006695466736952464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,6144,0.018605866034825645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,7168,0.01902186671892802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,5120,0.005914666752020518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,5120,0.016772266228993735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,6144,0.017427200078964235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,4096,0.005906133353710175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,4096,0.019447465737660728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,5120,0.01567893326282501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,3584,0.005514666438102722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,3584,0.017196800311406454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,4096,0.013699199755986533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,3072,0.0055402666330337524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,3072,0.018104533354441323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,3584,0.012853333353996277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,2560,0.005031466484069824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,2560,0.016683733463287352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,3072,0.011689600348472596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,2048,0.005485866467158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,2048,0.016357333461443583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,2560,0.010819199681282043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,1536,0.004434133569399515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,1536,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,2048,0.01004266639550527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,1024,0.003568000098069509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,1024,0.014627200365066529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,1536,0.009172266721725464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,768,0.003050666550795237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,768,0.014517333110173544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,1024,0.00871573289235433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,512,0.0030261332790056865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,512,0.014846932888031007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,768,0.007940266529719036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,256,0.0025962665677070618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,512,0.0075103998184204105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,256,0.014597333470980325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,256,0.007161599894364675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,128,0.002621866762638092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,128,0.014570666352907815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,128,0.007034666836261749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,64,0.002219733347495397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,64,0.014429866274197897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,1024,32,0.002219733347495397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,1024,32,0.014455466469128927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,65536,0.020718934138615926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,65536,0.03382506767908732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1536,65536,0.12403733730316162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,16384,0.010065066814422607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,16384,0.020679465929667153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,12288,0.009174399574597676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,12288,0.021652267376581828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,16384,0.034644265969594315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,10240,0.007708799839019775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,12288,0.028228267033894854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,10240,0.02034133275349935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,8192,0.006359466910362243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,8192,0.020786132415135702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,8192,0.020619734128316244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,7168,0.005921066800753275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,7168,0.018578133980433145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,10240,0.024498132864634196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,6144,0.006393600006898243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,6144,0.018812799453735353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,7168,0.018987733125686645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,5120,0.0058559998869895935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,5120,0.018326399723688762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,6144,0.01722666621208191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,4096,0.005987200140953064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,4096,0.017913599809010826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,1024,65536,0.12059840361277263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,5120,0.015377066532770791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,3584,0.006031999985376993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,3584,0.01655359963575999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,3072,0.005471999943256378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,3072,0.01699626644452413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,4096,0.01337493360042572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,2560,0.0054954667886098225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,2560,0.01567146678765615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,3584,0.01290880044301351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,3072,0.01162986656030019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,2048,0.004324266811211904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,2048,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,2560,0.011245866616566975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,1536,0.0042303999265035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,1536,0.01455893317858378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,1024,0.0034858666360378264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,1024,0.014548266927401224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,1536,0.009180800120035807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,768,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,768,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,1024,0.008384000261624653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,2048,0.009985066453615824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,512,0.002674133330583572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,512,0.015011200308799743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,256,0.0022986667851607008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,256,0.014552533626556396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,768,0.007913599908351897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,512,0.00759680022795995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,128,0.0022442666192849475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,128,0.014587733149528503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,64,0.002277333289384842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,256,0.0072405333320299785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,128,0.006351999938488007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,768,32,0.0022624000906944276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,64,0.014058666427930197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,65536,0.028065067529678345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,768,32,0.013845333456993103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,16384,0.007167999943097432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,65536,0.029496532678604127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,16384,0.02044373353322347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,12288,0.0069482664267222095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,12288,0.019654399156570433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,12288,0.027581866582234698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,10240,0.006305066744486491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,10240,0.01890346606572469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,16384,0.034745601812998454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,8192,0.006049066781997681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,8192,0.019388800859451293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,10240,0.024295467138290405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,7168,0.005463466544946035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,7168,0.018639999628067016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,8192,0.02069546580314636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,6144,0.006076799829800924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,6144,0.019396267334620156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,7168,0.01895466645558675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,5120,0.0054400001962979635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,5120,0.01781866749127706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,6144,0.017410133282343546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,4096,0.006188799937566122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,4096,0.018835200866063436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,5120,0.01566933294137319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,3584,0.005491200089454651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,3584,0.016586666305859886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,4096,0.013700266679128012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,3584,0.012426666418711345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,3072,0.005410133302211762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,3072,0.015482667088508605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,3072,0.011646933356920878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,2560,0.005018666883309682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,2560,0.015052800377209982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,768,65536,0.12093760172526043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,2048,0.0053151999910672505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,2048,0.01564800043900808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,2560,0.0107914666334788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,1536,0.00378560001651446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,1536,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,1024,0.003446399917205175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,2048,0.01037013332049052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,1024,0.014779733618100485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,1536,0.009136000275611877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,768,0.0032661333680152893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,768,0.014499200383822122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,512,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,512,0.014860799908638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,1024,0.008050133287906647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,256,0.0025685332715511323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,768,0.00812906672557195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,512,0.0074890668193499255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,256,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,128,0.0022154666483402254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,128,0.014465066790580749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,64,0.0021749332547187803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,256,0.007204266885916392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,128,0.006326400240262349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,512,32,0.0023039999107519788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,64,0.014434132973353067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,512,32,0.013411200046539307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,65536,0.016344533363978068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,16384,0.006780800223350525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,65536,0.026655999819437663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,16384,0.02085439960161845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,12288,0.005640533566474914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,12288,0.019139200448989868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,12288,0.028056534131368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,10240,0.005992533266544342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,10240,0.01909866730372111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,16384,0.035073065757751466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,8192,0.005982933441797892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,8192,0.0188810666402181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,10240,0.02478826642036438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,7168,0.005498666564623515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,8192,0.020623999834060668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,7168,0.018651733795801796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,6144,0.005207466582457224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,6144,0.019167999426523842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,7168,0.01913386583328247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,5120,0.005186133086681366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,6144,0.017428267002105712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,5120,0.01844053268432617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,4096,0.00476800004641215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,4096,0.017882666985193887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,5120,0.015846400459607442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,3584,0.005558399856090546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,4096,0.013692800203959146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,3584,0.016787199179331462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,3072,0.0051125332713127134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,3072,0.015123200416564942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,512,65536,0.12028266588846843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,2560,0.005241600175698599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,2560,0.016758400201797485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,3584,0.012915199995040894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,2048,0.00494400014479955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,2048,0.015799466768900552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,3072,0.012036266922950744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,1536,0.0038773333032925926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,1536,0.015773866573969522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,2560,0.01129813293615977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,1024,0.0030954666435718536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,2048,0.010030933221181234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,1024,0.014709333578745524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,1536,0.009574400385220845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,768,0.0030378667016824085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,768,0.014548266927401224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,512,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,1024,0.007969066500663757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,512,0.01402773360411326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,768,0.007930666704972585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,256,0.002204799900452296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,512,0.007153066496054332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,256,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,128,0.002284800012906392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,256,0.0073525334397951765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,128,0.014216533303260804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,64,0.002125866711139679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,128,0.006748799979686737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,256,32,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,64,0.013315199812253316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,65536,0.008778666456540424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,256,32,0.014086400469144186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,16384,0.005927466849486033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,65536,0.024717867374420166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,16384,0.020489599307378134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,12288,0.005443199972311656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,12288,0.019359999895095827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,16384,0.03495039939880371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,256,65536,0.11989759604136149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,10240,0.005512533088525137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,10240,0.01845653255780538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,12288,0.02764586607615153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,8192,0.005560533205668131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,10240,0.024284799893697105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,8192,0.016914133230845133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,7168,0.005345066885153452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,7168,0.017579734325408936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,6144,0.005710933109124502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,8192,0.02101866602897644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,6144,0.01886826753616333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,5120,0.0053247998158137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,7168,0.019052799542744955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,5120,0.01872640053431193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,4096,0.004886400202910105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,4096,0.019063466787338258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,6144,0.01730239987373352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,3584,0.004683733483155568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,5120,0.01574613352616628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,3584,0.016809600591659545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,3072,0.004636799792448679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,3072,0.015416533748308817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,4096,0.013828266660372415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,3584,0.012846933801968894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,2560,0.005195733408133189
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,2560,0.015094400445620219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,3072,0.011596799890200297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,2048,0.004384000102678935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,2048,0.014825600385665893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,1536,0.00359253336985906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,2560,0.010844799876213073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,1536,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,2048,0.009613866607348125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,65536,0.12033493518829345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,1024,0.002961066613594691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,768,0.0025909334421157838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,1024,0.0144896000623703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,1536,0.009308800101280212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,768,0.014439466595649719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,512,0.0026752000053723653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,1024,0.007999999821186066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,256,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,512,0.014332800110181173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,768,0.008026666442553202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,128,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,512,0.0075093333919843035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,256,0.014306132992108664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,256,0.00710399995247523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,64,0.0021578667064507803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,128,0.013989333311716715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,128,32,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,32,128,128,0.006698666512966156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,64,0.012971733013788858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,65536,0.008986666798591614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,128,32,0.012610133488972983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,12288,0.0053727999329566956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,16384,0.005876266459623972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,65536,0.02472426692644755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,16384,0.019590399662653604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,10240,0.004996266464392344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,8192,0.005512533088525137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,12288,0.018604799111684164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,10240,0.018504534165064493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,7168,0.005111466844876607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,8192,0.01730026602745056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,6144,0.005145599941412607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,7168,0.017142399152119955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,5120,0.004659200211366018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,6144,0.0185973326365153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,4096,0.004249600072701773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,5120,0.019717333714167277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,3584,0.005073066552480062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,4096,0.016660267114639284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,3072,0.004694400231043497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,3584,0.018329600493113198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,2560,0.004622933268547058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,3072,0.01608746647834778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,2048,0.004220800101757049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,2560,0.0164490669965744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,1536,0.0034677334129810332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,1024,0.003052799900372823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,2048,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,1536,0.014532267053922018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,768,0.0030026666820049284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,1024,0.014446933070818582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,512,0.002570666621128718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,256,0.002182399978240331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,768,0.013661866386731466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,128,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,512,0.014936533570289613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,256,0.01448853313922882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,64,0.0021664001047611236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,64,32,0.0018229333062966663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,64,0.013794133067131042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,128,0.013793067137400309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,65536,0.007457066575686138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,64,32,0.01446293294429779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,16384,0.005475200215975444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,12288,0.005028266708056132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,65536,0.024272000789642333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,16384,0.018525866667429607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,10240,0.005074133475621542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,12288,0.018525866667429607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,8192,0.004611200094223023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,10240,0.019859200716018675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,7168,0.0046517332394917805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,8192,0.018147200345993042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,6144,0.004690133531888326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,7168,0.018731733163197838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,5120,0.004701866706212362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,6144,0.016937599579493205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,4096,0.004268800218900045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,5120,0.01657600005467733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,3584,0.004655999938646952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,3072,0.004273066421349844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,4096,0.01727893352508545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,2560,0.004695466657479604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,3584,0.016544000307718913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,2048,0.004278400043646494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,3072,0.016596266627311708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,1536,0.003453866640726725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,2560,0.01616853376229604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,1024,0.002995199958483378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,2048,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,768,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,1536,0.0163263996442159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,512,0.0025418666501839954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,1024,0.014442666371663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,256,0.0021920000513394673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,768,0.01420906682809194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,128,0.0021749332547187803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,512,0.014406399925549826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,64,0.001961600035429001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,256,0.013771733641624451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,32,32,32,0.0021759999295075734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,128,0.014190933108329773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,64,0.01376106639703115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,32,32,32,0.014398933450380961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,16384,0.1794314702351888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,16384,0.3260554631551107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,12288,0.12533546288808187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,12288,0.21893760363260903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,10240,0.18325866063435872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,10240,0.10922453403472901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,8192,0.14763840039571124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,8192,0.08905813694000245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,8192,0.06661226749420165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,7168,0.1295786698659261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,7168,0.07973759969075521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,10240,0.08173440297444662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,6144,0.1118122657140096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,6144,0.07112853527069092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,12288,0.09856639703114828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,5120,0.09488853613535563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,5120,0.06286186774571736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,6144,0.05107413530349732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,4096,0.075272536277771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,4096,0.05321280161539713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,7168,0.061933867136637365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,3584,0.06664426724116007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,3584,0.04990293184916179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,5120,0.04493226607640584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,3072,0.05800960063934326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,3072,0.04330026706059774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,16384,0.12707520325978597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,2560,0.04863253434499105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,2560,0.038990934689839676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,3584,0.033369600772857666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,2048,0.03992533286412557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,2048,0.03544853528340657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,4096,0.036660265922546384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,1536,0.03094826738039653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,1536,0.030008532603581745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,3072,0.0301962673664093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,1024,0.022080000241597494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,1024,0.02479893366495768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,2560,0.026013867060343428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,768,0.017357865969340004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,768,0.022529067595799764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,1536,0.018652800718943277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,512,0.012317867080370585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,512,0.020549333095550536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,2048,0.022474666436513267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,256,0.007170133292675018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,256,0.016567466656366985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,256,0.009221333265304565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,128,0.004782933493455251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,128,0.01623679995536804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,128,0.0082997332016627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,64,0.004232533276081085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,64,0.015869866808255514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,65536,32,0.00421013335386912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,65536,32,0.01525973379611969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,512,0.011434666315714518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,65536,0.32924906412760413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,65536,0.1859231948852539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,16384,0.09143893718719483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,16384,0.055674668153127035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,768,0.013863466183344521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,12288,0.06905386447906495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,12288,0.04546773433685303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,65536,1024,0.015714133779207863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,10240,0.058841598033905027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,10240,0.04193173249562581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,10240,0.03202773332595825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,8192,0.042224001884460446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,8192,0.0357312003771464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,12288,0.034468265374501546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,7168,0.03692053159077962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,7168,0.033241599798202515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,16384,0.04569813410441081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,6144,0.0324618657430013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,6144,0.031591467062632245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,8192,0.026054400205612182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,5120,0.02733333309491475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,5120,0.027450666824976606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,7168,0.0238154669602712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,4096,0.022674133380254112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,4096,0.024765866994857787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,5120,0.018652800718943277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,3584,0.01967039903004964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,3584,0.025102933247884113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,6144,0.021500800053278604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,3072,0.017577600479125977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,3072,0.02265066703160604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,4096,0.016503467162450155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,2560,0.014972800016403198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,2560,0.020626133680343627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,2560,0.012885333100954691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,2048,0.012468266487121581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,2048,0.018948266903559365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,3584,0.01534293293952942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,1536,0.009597866733868917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,1536,0.018294399976730345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,3072,0.014057599504788718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,1024,0.007484800120194752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,1024,0.015983999768892924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,1024,0.008754133184750875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,768,0.005983999868233999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,768,0.014562132954597472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,1536,0.009975467125574749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,512,0.003399466723203659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,512,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,2048,0.011646933356920878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,256,0.00262719988822937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,256,0.014385066429773965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,512,0.007595733304818471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,128,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,128,0.01384213368097941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,128,0.006750933329264323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,64,0.0022079999248186748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,768,0.008370133241017659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,16384,32,0.002566399921973546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,64,0.01420266628265381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,16384,32,0.014523733655611673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,65536,0.24083733558654785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,65536,0.14530026117960612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,16384,0.06598186492919922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,16384,0.051173333326975504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,256,0.00705813318490982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,12288,0.048658132553100586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,12288,0.04118080139160156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,12288,0.029974399010340373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,10240,0.04189653396606445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,10240,0.03707306782404582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,16384,0.03800640106201172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,8192,0.03545920054117839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,8192,0.033369600772857666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,10240,0.026207999388376875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,7168,0.03099626700083415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,7168,0.030844799677530926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,8192,0.021934932470321654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,6144,0.027247999111811323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,6144,0.029526400566101074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,7168,0.02029973268508911
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,5120,0.023635200659434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,5120,0.023195733626683555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,6144,0.018209065993626913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,4096,0.01981546680132548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,4096,0.022826667626698813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,5120,0.016770132382710776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,3584,0.017771732807159425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,3584,0.020837332804997763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,4096,0.014498133460680643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,3072,0.015827199816703795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,3072,0.020705066124598184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,3584,0.013934933145840964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,2560,0.0116074671347936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,2560,0.02062826752662659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,3072,0.012479999661445617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,2048,0.009954133629798889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,2048,0.01930026610692342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,2560,0.0116266667842865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,1536,0.008457600076993307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,1536,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,1536,0.009236266215642292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,1024,0.006737066805362702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,1024,0.014754133423169455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,2048,0.010063999891281128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,768,0.005089066425959269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,768,0.014626133441925048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,768,0.007925333579381307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,512,0.003929600119590759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,512,0.01479680041472117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,1024,0.008406399687131246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,256,0.003068800022204717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,256,0.014589866995811463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,256,0.007147733370463054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,128,0.0025898667673269907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,128,0.014086400469144186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,512,0.0075541332364082335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,64,0.0026176000634829206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,64,0.0144405335187912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,12288,32,0.002611200014750163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,12288,32,0.014517333110173544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,65536,0.20325759251912437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,65536,0.13043839931488038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,128,0.006715733309586842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,16384,0.061426134904225674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,16384,0.04483306805292765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,16384,0.03738773266474406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,12288,0.044513066609700516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,12288,0.038412801424662274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,12288,0.029348266124725342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,10240,0.037020798524220785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,10240,0.035034668445587155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,16384,65536,0.16283413569132488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,8192,0.02977706591288249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,8192,0.03160853385925293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,10240,0.026016000906626386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,7168,0.027628799279530842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,7168,0.029049599170684816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,8192,0.021849600474039714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,6144,0.02435200015703837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,6144,0.028493867317835493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,7168,0.020286933581034342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,5120,0.020948266983032225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,5120,0.02537920077641805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,12288,65536,0.12983787059783936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,4096,0.017195733388264973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,4096,0.021100799242655434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,5120,0.016571733355522155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,3584,0.015726932883262636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,3584,0.020973867177963255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,6144,0.01830186645189921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,3072,0.013829333583513894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,3072,0.02057066758473714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,4096,0.014692266782124838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,2560,0.01134933332602183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,2560,0.01883946657180786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,3072,0.01237333317597707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,2048,0.00958720048268636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,2048,0.019718400637308755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,3584,0.013403733571370443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,1536,0.00831573357184728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,2560,0.011250133315722149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,1536,0.016697599490483602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,1024,0.006113066772619883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,1024,0.016150400042533875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,1024,0.008830933769543966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,768,0.0047978664437929785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,768,0.014800000190734863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,1536,0.009224533041318258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,512,0.004038399954636892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,512,0.014874666929244995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,768,0.008014933268229166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,256,0.003127466638882955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,256,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,512,0.0076117331782976794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,128,0.0026528000831604003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,128,0.013356799880663553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,2048,0.010016000270843506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,64,0.0022677332162857056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,64,0.01344000001748403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,10240,32,0.0025631998976071674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,256,0.007117866476376851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,10240,32,0.013690666357676188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,65536,0.1584885279337565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,128,0.0067562664548556015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,65536,0.10312000115712483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,16384,0.048970667521158855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,16384,0.03741439978281657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,12288,0.03483626842498779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,12288,0.03161493341128031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,12288,0.029254400730133058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,10240,0.03012053370475769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,10240,0.02829759915669759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,16384,0.036668801307678224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,8192,0.021843200922012328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,8192,0.024740266799926757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,10240,0.025618133942286174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,7168,0.019432532787322997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,7168,0.0230240007241567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,8192,0.021525333325068154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,6144,0.017422932386398315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,6144,0.022526933749516805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,7168,0.019949867328008016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,5120,0.01495573321978251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,10240,65536,0.12660480340321859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,5120,0.021826134125391642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,4096,0.012172800302505494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,4096,0.019024000565210978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,6144,0.0181877334912618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,3584,0.011121066411336263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,3584,0.020114133755366005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,4096,0.014069333672523499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,3072,0.009718400239944459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,3072,0.018606932957967122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,5120,0.016495999693870545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,2560,0.008752000331878663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,2560,0.01962453325589498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,3584,0.013018666704495748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,2048,0.008001066744327545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,2048,0.016635732849438985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,2560,0.011236266295115153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,1536,0.006808533271153767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,3072,0.012039466698964437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,1536,0.015217066804567973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,1024,0.004531200230121613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,1024,0.015659733613332113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,2048,0.010030933221181234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,768,0.003640533238649368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,768,0.0146506667137146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,768,0.008041599889596303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,1024,0.008431999882062276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,512,0.0030303999781608583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,512,0.014683733383814493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,256,0.002699733277161916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,256,0.0145087997118632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,1536,0.009191466371218364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,256,0.007167999943097432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,128,0.0022485333184401194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,128,0.014499200383822122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,64,0.0022656001150608064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,64,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,8192,32,0.002338133255640666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,8192,32,0.014500266313552857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,128,0.0067893331249554946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,65536,0.1460309346516927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,512,0.007546666761239369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,16384,0.03989973465601603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,65536,0.0932469367980957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,16384,0.03729386727015178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,12288,0.030033065875371294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,12288,0.028974932432174683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,12288,0.028810666004816694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,10240,0.02598080039024353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,10240,0.027426133553187054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,16384,0.03621439933776856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,8192,0.021473066012064616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,8192,0.02295573353767395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,10240,0.025174399216969807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,7168,0.019636267423629762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,7168,0.0228000005086263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,8192,0.02147093415260315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,6144,0.015646933515866598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,6144,0.023005867004394533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,7168,0.01989333430926005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,5120,0.013871999581654868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,5120,0.020829866329828896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,6144,0.01821440060933431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,4096,0.011132799585660299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,4096,0.018669867515563966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,5120,0.016499200463294984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,3584,0.010097066561381023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,3584,0.019793067375818887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,4096,0.014074666301409402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,3072,0.009164800246556599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,3072,0.018632533152898152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,3584,0.012984533111254373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,2560,0.008373333017031352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,2560,0.016720000902811685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,3072,0.011682132879892986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,2048,0.006985599795977275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,2048,0.017451733350753784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,2560,0.0108842670917511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,1536,0.006393600006898243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,1536,0.014656000336011252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,2048,0.010032000144322713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,1024,0.003903999924659729
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,1024,0.015383467078208923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,1536,0.009348266323407491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,768,0.0035317334036032355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,768,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,1024,0.008373333017031352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,512,0.0030421334008375804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,512,0.014634666840235391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,768,0.007955199976762136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,256,0.0027146667242050173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,256,0.014588800072669984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,512,0.007485866546630859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,128,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,128,0.013336533308029174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,128,0.006706133484840393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,64,0.0021578667064507803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,256,0.007175466914971669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,7168,32,0.0023605334262053173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,64,0.013505066434542337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,7168,32,0.013740799824396768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,65536,0.12622719605763752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,65536,0.08600107034047445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,16384,0.03520960013071696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,16384,0.034647464752197266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,8192,65536,0.1265226682027181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,12288,0.026492800315221148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,12288,0.030300800005594892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,16384,0.03547626733779907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,10240,0.023145600159962972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,10240,0.028090665737787884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,12288,0.028886399666468304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,8192,0.019409066438674925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,8192,0.022732800245285033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,10240,0.02518186569213867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,7168,0.017451733350753784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,7168,0.022402133544286093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,8192,0.021593600511550903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,6144,0.01588159998257955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,6144,0.020787199338277183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,7168,0.019556266069412232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,5120,0.013678933183352152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,5120,0.019476266702016194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,6144,0.018216532468795777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,4096,0.010391466816266378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,4096,0.020222934087117513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,5120,0.016155733664830526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,3584,0.00941973328590393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,3584,0.019214934110641478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,4096,0.01416000028451284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,3072,0.008583466211954754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,3072,0.01789120038350423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,3584,0.012472533186276754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,2560,0.007546666761239369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,2560,0.016212266683578492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,7168,65536,0.12586452960968017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,2048,0.006698666512966156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,3072,0.012177067001660664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,2048,0.016501333316167197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,1536,0.005542399982611338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,1536,0.01528320014476776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,2560,0.010795733332633973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,1024,0.0038624001046021783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,1024,0.014908799529075622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,1536,0.009192533294359843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,768,0.0035349334279696143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,768,0.014579199751218162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,2048,0.010427733262379963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,512,0.0031189332405726117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,512,0.014662399888038635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,768,0.00798826664686203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,256,0.0026880001028378804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,256,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,1024,0.008301866551240284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,128,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,128,0.013140267133712769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,256,0.007533866663773854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,64,0.0022869333624839783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,64,0.01420906682809194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,6144,32,0.0023391999304294584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,128,0.0070720002055168155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,512,0.007539199789365132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,6144,32,0.014060800274213156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,65536,0.10423146883646647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,16384,0.032758400837580366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,65536,0.0757536013921102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,16384,0.032433066765467325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,12288,0.02414399981498718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,12288,0.02884053389231364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,12288,0.028475733598073323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,10240,0.020579200983047486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,10240,0.026421332359313966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,16384,0.035947732130686444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,8192,0.01776533325513204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,8192,0.023496532440185548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,10240,0.025306665897369386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,7168,0.01585493286450704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,7168,0.022574933369954427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,8192,0.021566933393478392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,6144,0.014108799894650779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,6144,0.02037546634674072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,7168,0.019483733177185058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,5120,0.01204159955183665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,5120,0.018628267447153728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,6144,0.0178272008895874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,4096,0.009665066997210186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,4096,0.018758400281270345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,5120,0.015735466281572977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,3584,0.008706133564313252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,3584,0.017565866311391197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,4096,0.013767466942469279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,3072,0.007962666451931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,3072,0.01672853430112203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,3584,0.012474667032559712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,2560,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,2560,0.015828266739845276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,2560,0.010827733079592387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,2048,0.00544106662273407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,2048,0.015749333302179973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,3072,0.012200533350308736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,1536,0.0052597333987553915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,1536,0.015240533153216043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,2048,0.010020266969998677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,1024,0.003819733361403147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,1024,0.014670933286348978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,1536,0.00974079966545105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,768,0.0035082665582497918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,768,0.01461120049158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,1024,0.008328533172607422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,512,0.002995199958483378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,512,0.014775466918945313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,512,0.007477333148320516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,256,0.0026026666164398193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,6144,65536,0.12573760350545246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,256,0.014513066411018372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,128,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,128,0.01368106702963511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,768,0.008040533463160197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,64,0.0023530667026837667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,64,0.0134442667166392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,5120,32,0.0022976001103719073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,128,0.006750933329264323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,5120,32,0.014364799857139588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,65536,0.0840949296951294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,256,0.006937600175539653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,16384,0.02626986702283223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,65536,0.062259201208750406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,16384,0.02826559940973918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,12288,0.01991893251736959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,12288,0.025141332546869916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,12288,0.0284661332766215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,10240,0.016829866170883178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,10240,0.02309653361638387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,16384,0.03580160140991211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,8192,0.01399679978688558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,8192,0.02469653288523356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,10240,0.025501867135365803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,7168,0.012843733032544455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,7168,0.023678932587305704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,8192,0.021079466740290324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,6144,0.011641599734624227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,6144,0.022407466173172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,7168,0.019038933515548705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,5120,0.010517332951227825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,5120,0.01867413322130839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,6144,0.01744106610616048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,4096,0.012187733252843221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,4096,0.01657600005467733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,5120,0.01570026675860087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,3584,0.010441600282986959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,3584,0.017064533631006875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,4096,0.014115200440088908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,3072,0.009594666957855224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,3072,0.015089066823323569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,3584,0.012458667159080505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,2560,0.005957333246866862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,2560,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,3072,0.012468266487121581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,2048,0.004996266464392344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,2048,0.015492266416549683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,2560,0.010860799749692281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,1536,0.003874133278926214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,1536,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,2048,0.00998293360074361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,1024,0.0034133332471052804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,1024,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,1536,0.009161600470542907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,768,0.003005866706371307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,768,0.014486400286356607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,1024,0.008797867099444072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,512,0.0026496000587940215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,512,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,768,0.007919999957084655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,256,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,256,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,512,0.0074976002176602675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,128,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,128,0.013825066884358726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,256,0.007110400001207988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,64,0.002195200075705846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,128,0.0071168000499407455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,64,0.013411200046539307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,4096,32,0.0022570667167504626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,4096,32,0.013649066289265951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,65536,0.07248746554056804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,65536,0.05684693257013956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,16384,0.022729599475860597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,16384,0.027772800127665205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,5120,65536,0.12520853678385419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,12288,0.017410133282343546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,12288,0.025010132789611818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,12288,0.028700800736745198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,10240,0.015292800466219583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,10240,0.022989867130915324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,16384,0.03562026818593343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,8192,0.013022933403650919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,8192,0.024950400988260905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,8192,0.020680532852808634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,7168,0.012065066893895467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,7168,0.022613332668940226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,10240,0.025131734212239583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,6144,0.010998400052388509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,6144,0.021792000532150267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,7168,0.018986666202545167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,5120,0.010033067067464192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,5120,0.01841813325881958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,6144,0.01734293301900228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,4096,0.01162453293800354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,4096,0.017014400164286295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,4096,0.013690666357676188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,3584,0.010426666339238484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,3584,0.016679465770721436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,5120,0.015780267119407655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,3072,0.009495466947555542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,3072,0.016436266899108886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,4096,65536,0.12513919671376544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,2560,0.005485866467158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,2560,0.014665599664052328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,3072,0.01209920048713684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,2048,0.0047775998711586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,2048,0.015435733397801719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,3584,0.012549333771069846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,1536,0.0039434666434923805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,1536,0.014564266800880432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,2560,0.010886399944623312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,1024,0.0034400001168251038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,1024,0.014500266313552857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,1536,0.00923413336277008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,768,0.003176533430814743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,768,0.014547200004259745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,2048,0.010052266716957092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,512,0.002739199995994568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,512,0.014853333433469137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,1024,0.008400000135103862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,256,0.0025610665480295816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,256,0.01442026694615682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,768,0.007997866471608479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,128,0.0022581333915392557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,128,0.01346560021241506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,256,0.007089066505432129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,64,0.002181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,128,0.006727466483910878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,64,0.013320533434549966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3584,32,0.0022367998957633973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3584,32,0.014102400342623393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,65536,0.06342080036799112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,512,0.007567999760309856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,65536,0.054372266928354895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,16384,0.01986560026804606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,16384,0.02684906721115112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,12288,0.015307733416557312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,12288,0.02406826615333557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,12288,0.028068266312281293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,10240,0.013717333475748697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,10240,0.023142399390538533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,16384,0.035930665334065755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,8192,0.012156800429026286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,8192,0.021669334173202513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,10240,0.02476053237915039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,7168,0.011096533139546711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,7168,0.020652800798416138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,8192,0.0208512008190155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,6144,0.010502400000890096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,6144,0.020853332678476968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,7168,0.018974934021631876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,5120,0.00922986666361491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,5120,0.01853546698888143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,6144,0.017541333039601644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,4096,0.008612266182899475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,4096,0.01676373283068339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,5120,0.015659733613332113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,3584,0.00927786628405253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,3584,0.016596266627311708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,4096,0.01378666659196218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,3584,0.012874666849772134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,3072,0.008605866630872091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,3072,0.01544319987297058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,2560,0.005356800059477488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,2560,0.0146506667137146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3584,65536,0.12555306752522785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,2048,0.004578133424123129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,2048,0.014620799819628397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,2560,0.010757333040237427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,1536,0.003819733361403147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,1536,0.01448853313922882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,3072,0.011687466502189636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,1024,0.00348693331082662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,1024,0.015734400351842245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,1536,0.009105066458384197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,768,0.003011200080315272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,768,0.014114133516947427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,2048,0.010061867038408915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,512,0.0027445333699385325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,1024,0.008109866579373678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,512,0.014349866906801859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,256,0.0026837334036827086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,256,0.014273066322008768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,768,0.007922133306662242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,256,0.007130666573842366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,128,0.002204799900452296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,128,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,512,0.007539199789365132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,64,0.002183466653029124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,64,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,3072,32,0.0021589333812395734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,128,0.0067456002036730455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,3072,32,0.01455573340257009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,65536,0.0542794664700826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,65536,0.0494271993637085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,16384,0.01813973387082418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,16384,0.02609493335088094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,12288,0.013540266950925191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,12288,0.02287893295288086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,12288,0.028122667471567792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,10240,0.012477866808573405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,10240,0.02079040010770162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,16384,0.03596160014470418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,8192,0.010752000411351522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,8192,0.021259733041127524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,10240,0.024596265951792397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,7168,0.009955199559529622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,7168,0.020733867088953653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,8192,0.020814933379491172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,6144,0.009388800462086995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,6144,0.020297600825627645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,7168,0.01865066687266032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,5120,0.008403199911117553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,5120,0.01949866612752279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,6144,0.017488000790278117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,4096,0.007294933497905731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,4096,0.017270400126775106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,5120,0.015753600001335143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,3584,0.008409600456555684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,3584,0.016613333423932394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,4096,0.013806933164596557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,3072,0.007503999769687653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,3072,0.016693333784739174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,3584,0.012488533059755962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,2560,0.005086933573087057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,2560,0.014550399780273438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,3072,0.011715199549992878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,2048,0.004716800153255462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,2048,0.015502933661142984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,2560,0.010824533303578694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,1536,0.003861333429813385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,1536,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,1536,0.009172266721725464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,1024,0.0033301333586374915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,1024,0.015313067038853965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,2048,0.009993599851926167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,768,0.00310506671667099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,768,0.01461120049158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,3072,65536,0.1251893361409505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,512,0.002833066632350286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,1024,0.007925333579381307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,512,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,256,0.0026837334036827086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,768,0.00803413341442744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,256,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,128,0.0022805333137512207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,128,0.013670399785041809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,512,0.007499733567237854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,64,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,256,0.007190399865309398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,128,0.006855466465155284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2560,32,0.00223786657055219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,64,0.013009066383043924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,65536,0.04384746551513672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2560,32,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,65536,0.044394667943318686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,16384,0.015924266974131265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,16384,0.022738132874170938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,12288,0.011424000064531963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,12288,0.02536639968554179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,12288,0.02813120086987813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,10240,0.010368000467618306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,10240,0.022796799739201866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,16384,0.035607465108235675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,8192,0.009352533022562663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,8192,0.02107200026512146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,10240,0.02474986712137858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,7168,0.008801066875457763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,7168,0.020787199338277183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,8192,0.02090453306833903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,6144,0.011178666353225708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,6144,0.01904426614443461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,7168,0.0190720001856486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,5120,0.00993280013402303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,5120,0.01689280072848002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,6144,0.017130666971206666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,4096,0.008539733290672303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,4096,0.01697173317273458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,5120,0.01573973298072815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,3584,0.0076447998483975725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,3584,0.016796799500783284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,3584,0.012458667159080505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,3072,0.005850666761398315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,3072,0.014923733472824097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,4096,0.013773866494496665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,2560,0.0051701332132021586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,3072,0.011610666910807293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,2560,0.01637440025806427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,2048,0.004756266872088114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,2048,0.01667733391125997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2560,65536,0.1252842664718628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,1536,0.00391146664818128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,1536,0.014566399653752646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,1536,0.00920960009098053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,1024,0.003432533393303553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,1024,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,2560,0.010805333654085796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,768,0.0030229332546393077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,768,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,2048,0.010029866298039754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,512,0.0026634665826956432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,512,0.014614400267601014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,768,0.007570133109887441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,256,0.0023520000278949736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,256,0.01448853313922882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,512,0.007276799778143566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,128,0.0022528000175952912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,1024,0.008348799745241801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,128,0.01439466675122579
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,64,0.0021546666820844015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,64,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,128,0.007162666817506154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,256,0.007090133428573608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,2048,32,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,65536,0.03451093435287476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,2048,32,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,16384,0.017810134092966716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,65536,0.03770453135172526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,16384,0.02648959954579671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,12288,0.01460693379243215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,12288,0.02373866637547811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,12288,0.028124799331029255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,10240,0.012947199741999307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,10240,0.020651733875274657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,16384,0.03508906761805217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,8192,0.01125759979089101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,8192,0.020322134097417198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,10240,0.024378667275110878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,7168,0.010445866982142131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,7168,0.021916800737380983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,8192,0.020705066124598184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,6144,0.009569066762924194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,6144,0.01814080079396566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,7168,0.01954453388849894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,5120,0.00881173312664032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,5120,0.016770132382710776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,6144,0.01741973360379537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,4096,0.007236266632874806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,4096,0.017064533631006875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,5120,0.015718400478363037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,3584,0.0067104001839955645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,3584,0.016985599199930826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,4096,0.013636266191800436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,3072,0.005568000177542368
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,3072,0.017079466581344606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,3584,0.012863999605178833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,2560,0.005038933455944061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,2560,0.01590506633122762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,3072,0.011566932996114094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,2048,0.005962666869163513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,2048,0.01627413332462311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,2560,0.011225600043932598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,1536,0.0038581334054470064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,1536,0.014594133694966635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,2048,0.00997440020243327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,1024,0.0033930666744709016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,1024,0.01462399959564209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,1536,0.009159466624259949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,768,0.0030346666773160298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,768,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,1024,0.008025600016117096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,512,0.0026549334327379864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,512,0.01472106675306956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,2048,65536,0.12455146312713623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,256,0.0026208000878492994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,256,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,768,0.00788800021012624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,128,0.0021898667017618816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,512,0.007469866673151653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,128,0.014451199769973755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,64,0.002181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,64,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,256,0.0071285332242647815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1536,32,0.0021930667261282604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,128,0.006692266464233399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1536,32,0.013324800133705138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,65536,0.024800000588099162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,16384,0.013765333096186319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,65536,0.0309663991133372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,16384,0.021401600042978922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,12288,0.011242666840553283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,12288,0.019937066237131755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,12288,0.028092799584070842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,10240,0.01002239982287089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,10240,0.02044373353322347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,16384,0.03504853248596192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,8192,0.008551466464996337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,8192,0.019293866554896035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,10240,0.024370133876800537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,7168,0.007658666869004567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,7168,0.018594133853912353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,7168,0.019402666886647543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,6144,0.007085866729418437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,6144,0.01761173407236735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,8192,0.021052799622217813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,5120,0.006562133133411407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,5120,0.018292266130447387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,5120,0.015727999806404113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,4096,0.00610453337430954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,4096,0.01814080079396566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,6144,0.01739413340886434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,3584,0.0058890665570894875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,3584,0.016666666666666666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,3584,0.012888532876968384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,3072,0.005470933516820272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,4096,0.013689600427945457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,3072,0.016622933745384216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,2560,0.005004799862702688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,2560,0.017064533631006875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,2560,0.010867200295130412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,2048,0.006300800045331319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,3072,0.011711999773979187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,2048,0.016284799575805663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,1536,0.005116799970467886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,1536,0.01518933375676473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,1536,0.009213866790135701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,1024,0.0042303999265035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,1024,0.014680533607800802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,2048,0.00995733340581258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,768,0.003509333233038584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,768,0.014572800199190775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,1024,0.008348799745241801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,512,0.003009066730737686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,768,0.007891199986139933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,512,0.014437333742777506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,256,0.0025962665677070618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,256,0.014237866799036662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,512,0.007838933169841767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,128,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1536,65536,0.12231252988179524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,256,0.007109333574771881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,64,0.002181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,128,0.014433067043622336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,1024,32,0.0022240000466505687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,64,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,1024,32,0.014429866274197897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,65536,0.020360533396402994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,128,0.006278400123119354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,16384,0.011691733201344808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,65536,0.034842665990193686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,16384,0.02160960038503011
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,12288,0.009354666868845622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,12288,0.020558933417002358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,12288,0.02772480050722758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,10240,0.008331733445326488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,10240,0.019036799669265747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,16384,0.035173332691192626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,8192,0.007288533449172974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,8192,0.018973867098490395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,10240,0.024382932980855306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,7168,0.006625066697597504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,7168,0.018888533115386963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,8192,0.02076693375905355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,6144,0.006776533524195353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,6144,0.018590933084487914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,7168,0.01895786722501119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,5120,0.0062720000743865965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,5120,0.017102932929992674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,6144,0.017406932512919106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,4096,0.006093866626421611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,4096,0.01767680048942566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,5120,0.015292800466219583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,3584,0.005806933343410492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,3584,0.016966400543848674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,1024,65536,0.12076053619384766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,3072,0.005760000149408976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,3072,0.016029866536458333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,3584,0.01246506671110789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,4096,0.013327999909718832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,2560,0.005446400245030721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,2560,0.01685546636581421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,2048,0.004374399781227112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,2048,0.016364799936612447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,3072,0.01170133352279663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,1536,0.004300799965858459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,1536,0.014601600170135499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,2560,0.010750933488210043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,1024,0.0037823999921480812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,1024,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,2048,0.010029866298039754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,768,0.0035103999078273775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,768,0.01455893317858378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,1536,0.0091839998960495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,1024,0.007922133306662242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,512,0.003081600119670232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,512,0.014689067006111145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,768,0.007645866771539052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,256,0.002611200014750163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,256,0.01455466647942861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,512,0.007566933333873749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,128,0.0022069332500298817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,128,0.013370666901270548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,256,0.007143466671307881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,64,0.0022495999932289124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,64,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,128,0.006676266590754191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,768,32,0.0022602667411168414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,65536,0.028552534182866414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,768,32,0.013251200318336487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,65536,0.029973334074020384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,16384,0.008250666658083598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,16384,0.020383999745051066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,12288,0.0076789334416389465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,12288,0.020092799266179403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,16384,0.03482986688613891
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,10240,0.007302399973074596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,12288,0.028034132719039918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,10240,0.018758400281270345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,8192,0.006471466521422069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,8192,0.019291732708613077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,10240,0.02418559988339742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,8192,0.020670932531356812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,7168,0.005468800167242686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,7168,0.018823466698328652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,6144,0.006309333443641663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,6144,0.019008000691731773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,7168,0.01902186671892802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,5120,0.005834666887919108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,5120,0.017373865842819212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,6144,0.01704746683438619
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,4096,0.005868799984455109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,4096,0.017118932803471883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,5120,0.01569386621316274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,3584,0.005481599768002828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,3584,0.01680213411649068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,4096,0.013287466764450074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,3072,0.005504000186920166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,3072,0.016266666849454246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,3584,0.012812800208727517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,768,65536,0.12092586358388265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,2560,0.005037866532802582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,3072,0.01202453374862671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,2560,0.015223466356595359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,2048,0.004827733337879181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,2048,0.014697600404421488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,1536,0.003874133278926214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,1536,0.014604799946149192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,2560,0.010376532872517902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,1024,0.003010133405526479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,2048,0.009991466999053955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,1024,0.01467519998550415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,768,0.002995199958483378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,1536,0.009170132875442504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,768,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,512,0.0026549334327379864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,512,0.014703999956448874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,1024,0.007855999966462452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,768,0.00837546686331431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,256,0.0025898667673269907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,256,0.013885866602261862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,512,0.00746559997399648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,128,0.0022410665949185687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,256,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,128,0.013378133376439413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,64,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,512,32,0.0021759999295075734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,128,0.006762666503588359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,64,0.01333440045515696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,65536,0.016434133052825928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,512,32,0.014432000120480857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,16384,0.007121066749095917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,65536,0.02595199942588806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,16384,0.01971093416213989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,12288,0.005794133245944977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,12288,0.018824533621470133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,12288,0.027778132756551104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,10240,0.005895466605822245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,16384,0.03457066615422567
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,10240,0.0204202671845754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,8192,0.005619200070699056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,8192,0.017658666769663493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,10240,0.024332799514134727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,7168,0.005452799797058106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,8192,0.020320000251134236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,7168,0.018691200017929076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,6144,0.005093333125114441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,6144,0.01663040022055308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,6144,0.01732586622238159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,7168,0.018952532609303793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,5120,0.0050687998533248905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,5120,0.018824533621470133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,4096,0.004794666667779287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,4096,0.017819732427597046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,5120,0.015281066298484802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,3584,0.005036800106366476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,3584,0.01746986707051595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,4096,0.013307733337084451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,3584,0.012863999605178833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,3072,0.004922666649023692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,512,65536,0.12041386763254802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,3072,0.016515200336774193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,2560,0.005162666738033295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,2560,0.015313067038853965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,3072,0.011766399939854939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,2048,0.004683733483155568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,2560,0.0107914666334788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,2048,0.01597760021686554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,1536,0.003824000060558319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,1536,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,2048,0.010007466872533162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,1024,0.0030858665704727173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,1536,0.009475200374921163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,1024,0.01537493367989858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,768,0.0030421334008375804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,768,0.014526933431625366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,1024,0.007969066500663757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,512,0.0028597332537174227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,512,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,768,0.007573333382606506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,256,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,256,0.014441600441932679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,128,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,512,0.0072202667593956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,256,0.0067775999506314594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,64,0.002164266755183538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,128,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,256,32,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,64,0.013352533181508383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,256,32,0.014106667041778565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,128,0.006833066542943318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,65536,0.009554133812586466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,16384,0.005515733361244201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,65536,0.023601067066192628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,16384,0.01877440015474955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,12288,0.005178666611512502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,12288,0.018605866034825645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,12288,0.02800426681836446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,10240,0.005449600021044413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,16384,0.03460799853006999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,256,65536,0.12028373082478841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,10240,0.018696532646814982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,8192,0.00561706672112147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,8192,0.017851734161376955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,7168,0.005123200019200643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,7168,0.018595200777053834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,8192,0.020678400993347168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,10240,0.02428800066312154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,6144,0.005458133419354757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,6144,0.018333866198857626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,5120,0.004682666560014089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,7168,0.018959999084472656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,5120,0.019831466674804687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,4096,0.00461760014295578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,4096,0.017540266116460167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,6144,0.017539199193318686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,5120,0.01532586713631948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,3584,0.005052799979845682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,3584,0.01725013256072998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,3072,0.00421973317861557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,4096,0.013226667046546936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,3072,0.014541866381963095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,2560,0.004649599889914194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,3584,0.012880000472068786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,2560,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,2048,0.004248533149560293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,3072,0.011557333668073018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,2048,0.016266666849454246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,1536,0.003470933437347412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,2560,0.011082667112350463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,1536,0.01446293294429779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,1024,0.0029845332105954488
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,2048,0.010033067067464192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,1024,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,768,0.0026176000634829206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,1536,0.009567999839782714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,768,0.013268267114957174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,1024,0.007896533111731212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,512,0.00223786657055219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,768,0.007926400005817413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,512,0.014636799693107605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,256,0.0022272000710169475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,512,0.0075328002373377485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,256,0.014427733421325684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,128,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,256,0.007119999825954437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,128,0.014494933684666953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,64,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,128,0.006324266890684764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,128,32,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,64,0.014113066593805948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,65536,0.009187199672063192
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,128,32,0.012470400333404541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,16384,0.005508266886075338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,12288,0.005048533280690512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,65536,0.023164800802866616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,10240,0.005062399804592133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,16384,0.019272534052530925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,12288,0.01790720025698344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,8192,0.0054666668176651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,10240,0.016657066345214844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,7168,0.005052799979845682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,8192,0.01921280026435852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,6144,0.005058133105436961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,7168,0.017604267597198485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,5120,0.004203733305136363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,6144,0.016561067104339598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,5120,0.016960000991821288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,4096,0.00462719996770223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,3584,0.005087999999523163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,16,128,65536,0.11944533189137777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,4096,0.01657386620839437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,3072,0.004285866518815359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,3584,0.0165994664033254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,2560,0.004635733366012573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,3072,0.015075199802716575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,2048,0.004285866518815359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,2560,0.016819200913111367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,1536,0.003457066665093104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,2048,0.014607999722162882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,1024,0.003013333429892858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,1536,0.015642666816711427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,768,0.002621866762638092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,512,0.002219733347495397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,1024,0.01452906628449758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,768,0.014599466323852539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,256,0.002221866697072983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,128,0.0021920000513394673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,512,0.014382933576901754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,256,0.01406826674938202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,64,0.0018144000321626663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,64,32,0.0018346666047970455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,128,0.013146666685740152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,64,0.014350933829943338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,65536,0.007448533177375793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,64,32,0.012640000383059184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,16384,0.005126399795214335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,12288,0.005083733300367991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,65536,0.021615999937057494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,16384,0.018041600783665977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,10240,0.005098666747411092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,8192,0.004717866579691568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,12288,0.018655999501546224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,10240,0.01665173371632894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,7168,0.004711466530958811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,6144,0.0046304002404212955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,8192,0.017432532707850137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,7168,0.01795413295427958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,5120,0.004664533336957296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,4096,0.004299733539422353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,6144,0.01790613333384196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,5120,0.017333332697550455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,3584,0.004658133288224538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,3072,0.004266666869322458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,4096,0.0166101336479187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,3584,0.017298134167989095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,2560,0.00469760000705719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,2048,0.004660266637802124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,3072,0.014905599753061929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,2560,0.01448853313922882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,1536,0.003398400048414866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,1024,0.0030218665798505146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,2048,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,1536,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,768,0.002648533384005229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,512,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,1024,0.014469333489735923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,256,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,768,0.013384532928466798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,128,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,512,0.014149333039919535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,64,0.0022282667458057405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,256,0.014382933576901754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,128,0.012621866663297019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,16,32,32,0.001844266677896182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,64,0.014421332875887552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,16,32,32,0.012591999769210816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,16384,0.32730134328206384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,16384,0.17981972694396972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,12288,0.2417247931162516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,12288,0.14242134094238282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,10240,0.20303786595662437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,10240,0.10786133607228596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,8192,0.14667627016703289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,8192,0.08753173351287842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,8192,0.07139413356781006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,7168,0.1288319985071818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,7168,0.08034239610036215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,10240,0.08055573304494222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,6144,0.11103146870930988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,6144,0.06991146405537924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,12288,0.09938773314158121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,5120,0.09389440218607584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,5120,0.06162453492482504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,6144,0.05089813470840454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,4096,0.07499840259552001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,4096,0.0528767983118693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,7168,0.06131199995676676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,3584,0.06607360045115153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,3584,0.04875306685765584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,5120,0.045110400517781576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,3072,0.057144534587860105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,3072,0.0434933344523112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,16384,0.13389973640441893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,2560,0.04841599861780803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,2560,0.038747731844584146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,3584,0.03457173506418864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,2048,0.039426132043202715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,2048,0.03444266716639201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,4096,0.036157866319020585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,1536,0.030509867270787555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,1536,0.02949013312657674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,2560,0.027900799115498857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,1024,0.021861332654953002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,1024,0.024771199623743693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,3072,0.030430932839711506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,768,0.017437867323557534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,768,0.02240426739056905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,2048,0.0239519993464152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,512,0.012504532933235168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,1536,0.019333332777023315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,512,0.019326933224995933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,256,0.00718506673971812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,256,0.016525866587956746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,1024,0.015476266543070475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,128,0.004735999802748362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,128,0.015639467040697734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,768,0.014038399855295817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,64,0.004328533510367076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,64,0.01670080025990804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,65536,32,0.004411733150482178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,65536,32,0.016615466276804606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,65536,0.3226698557535807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,65536,0.18340907096862794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,128,0.008366933465003968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,16384,0.09424106280008951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,16384,0.06169066826502482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,256,0.009217066566149394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,12288,0.0683072010676066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,12288,0.043347199757893876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,65536,512,0.010759466886520385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,10240,0.05800426801045736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,10240,0.03954133192698161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,10240,0.03180906573931376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,8192,0.048077865441640215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,8192,0.034917334715525314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,12288,0.03822933435440064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,7168,0.03537813425064087
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,7168,0.03207040031750997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,16384,0.05044693152109782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,6144,0.031173332532246904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,6144,0.028971733649571736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,8192,0.02699626684188843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,5120,0.026562132438023883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,5120,0.027831466992696126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,7168,0.024987733364105223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,4096,0.02190613349278768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,4096,0.024596265951792397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,6144,0.022465066115061442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,3584,0.019358932971954346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,3584,0.02281493345896403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,4096,0.017093332608540852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,3072,0.017156267166137697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,3072,0.020848000049591066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,5120,0.01989439924558004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,2560,0.01458346645037333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,2560,0.020823466777801513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,2560,0.012883200248082479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,2048,0.012460800011952718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,2048,0.019606399536132812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,3584,0.0160480002562205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,1536,0.009565866986910502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,1536,0.018076799313227334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,3072,0.014660267035166421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,1024,0.0073632001876831055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,1024,0.014725333452224732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,1024,0.00881173312664032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,768,0.005987200140953064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,768,0.014637866616249084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,1536,0.010195199648539226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,512,0.0034143999218940735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,512,0.01444906691710154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,768,0.00843519965807597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,256,0.0026752000053723653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,512,0.0075647999842961625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,256,0.014132266243298849
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,128,0.0023413332800070446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,128,0.013063466548919678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,128,0.007214933137098948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,64,0.002221866697072983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,2048,0.011685333649317424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,64,0.013784533739089966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,16384,32,0.0022677332162857056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,16384,32,0.014392532904942832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,65536,0.2357482592264811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,65536,0.14294080734252929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,256,0.007503999769687653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,16384,0.0705514669418335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,16384,0.05010453462600708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,12288,0.048648532231648764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,12288,0.04166933298110962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,12288,0.02998080054918925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,10240,0.0420085350672404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,10240,0.03777279853820801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,16384,0.03811093171437581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,8192,0.03493653138478597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,8192,0.032519465684890746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,10240,0.025973333915074663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,7168,0.03095253308614095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,7168,0.03158400058746338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,8192,0.022155733903249104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,6144,0.026842667659123735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,6144,0.0276256004969279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,7168,0.020168532927831016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,5120,0.02349546750386556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,5120,0.026200532913208008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,6144,0.01845546762148539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,4096,0.01945599913597107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,4096,0.02487679918607076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,5120,0.01657919983069102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,3584,0.017465599377950034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,3584,0.02086506684621175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,4096,0.014563199877738953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,3072,0.01566933294137319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,3072,0.019497599204381308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,3584,0.013510400056838989
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,2560,0.013595733046531677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,2560,0.018658133347829182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,3072,0.01279039978981018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,2048,0.011989333232243856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,2048,0.018357332547505698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,2560,0.011236266295115153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,1536,0.008044800162315369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,1536,0.015434666474660238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,2048,0.010231467088063557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,1024,0.00639573335647583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,1024,0.014978133638699851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,1536,0.009215999643007915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,768,0.00418453315893809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,768,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,768,0.007886933286984761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,512,0.003045333425203959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,512,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,1024,0.008444799979527792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,256,0.0026954665780067446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,256,0.013501866658528646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,512,0.00755626658598582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,128,0.0022367998957633973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,128,0.012829867005348206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,256,0.007177599767843883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,64,0.0022645334402720133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,64,0.013917866349220275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,12288,32,0.0023103999594847363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,128,0.007095466554164887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,12288,32,0.013485866785049438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,65536,0.1983402729034424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,65536,0.12645546595255536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,16384,0.06118186712265015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,16384,0.04296213388442993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,16384,0.03720320065816243
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,12288,0.041262932618459064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,12288,0.03769493500391642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,12288,0.029227733612060547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,10240,0.03581546545028687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,10240,0.03463360071182251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,16384,65536,0.15397653579711915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,8192,0.02906773289044698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,8192,0.029496532678604127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,10240,0.02513813376426697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,7168,0.02683626612027486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,7168,0.028988800446192425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,12288,65536,0.1320138692855835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,6144,0.023784534136454264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,6144,0.026884265740712482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,8192,0.021941334009170532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,5120,0.02033920089403788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,5120,0.02637333273887634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,7168,0.019811199108759562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,4096,0.016571733355522155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,4096,0.02225066622098287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,4096,0.014494933684666953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,3584,0.015412267049153647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,3584,0.023433599869410196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,5120,0.016873600085576375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,3072,0.013777066270510355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,3072,0.020603734254837035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,6144,0.018262400229771932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,2560,0.012493866682052612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,2560,0.01990933418273926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,2560,0.01079360047976176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,2048,0.009638399879137675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,2048,0.01899519960085551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,3072,0.01245013376077016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,1536,0.007589333256085714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,1536,0.01652906636397044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,3584,0.013630933562914529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,1024,0.005971199770768484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,1024,0.014664533734321594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,1536,0.00916373332341512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,768,0.004735999802748362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,768,0.014413866400718688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,1024,0.008313600222269695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,512,0.00420906643072764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,512,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,2048,0.010036266843477885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,256,0.0026357332865397137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,256,0.014442666371663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,768,0.007993599772453308
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,128,0.0022272000710169475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,128,0.014451199769973755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,256,0.007117866476376851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,64,0.002218666672706604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,512,0.00749120016892751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,10240,32,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,64,0.01434346636136373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,10240,32,0.013336533308029174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,65536,0.15731627146402996
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,65536,0.09959572950998942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,16384,0.04871466557184855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,128,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,16384,0.03467520078023274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,12288,0.03395093282063802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,12288,0.029972267150878907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,12288,0.02886613408724467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,10240,0.029433600107828778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,10240,0.028424533208211263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,16384,0.036660265922546384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,8192,0.021868799130121866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,8192,0.025251199801762898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,10240,0.025441066424051924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,7168,0.0194048007329305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,7168,0.023703465859095253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,8192,0.02185279925664266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,6144,0.0173632005850474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,6144,0.02267626722653707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,7168,0.019825067122777304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,5120,0.014893866578737893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,5120,0.020614399512608846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,10240,65536,0.12651413281758625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,4096,0.012032000223795573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,4096,0.020811732610066733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,6144,0.018279467026392618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,3584,0.01100373367468516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,3584,0.01872106591860453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,5120,0.016438399751981102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,3072,0.009659733374913533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,3072,0.018603734175364175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,4096,0.014095999797185264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,2560,0.008755200107892354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,2560,0.016615466276804606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,3584,0.013324800133705138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,2048,0.007483733197053273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,2048,0.014906666676203408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,2560,0.010854400197664897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,1536,0.006284800171852112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,1536,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,3072,0.011803733309110005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,1024,0.00395413339138031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,1024,0.014588800072669984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,2048,0.010102400183677673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,768,0.0032469332218170166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,768,0.014642133315404256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,768,0.008113066852092742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,1536,0.009201066692670186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,512,0.0026805333793163298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,1024,0.008353066444396973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,256,0.002639999985694885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,512,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,256,0.013702399532000222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,128,0.002328533430894216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,128,0.013471999764442444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,128,0.006900266806284587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,64,0.002250666668017705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,256,0.006691200037797292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,64,0.013425067067146301
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,8192,32,0.0021429332594076794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,8192,32,0.013838932911554972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,65536,0.13943999608357746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,65536,0.09006933371225992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,16384,0.04010879993438721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,16384,0.03664746681849162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,512,0.00761599987745285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,12288,0.02942933241526286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,12288,0.02786239981651306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,12288,0.028860799471537274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,10240,0.025549866755803424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,10240,0.02579306761423747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,16384,0.03588053385416667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,8192,0.021079466740290324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,8192,0.022734934091567995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,10240,0.025576533873875935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,7168,0.018999467293421425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,7168,0.02228800058364868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,8192,0.02184213399887085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,6144,0.01579200029373169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,6144,0.0206389327843984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,7168,0.020122667153676353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,5120,0.013757866621017457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,5120,0.021079466740290324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,6144,0.018180267016092936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,4096,0.011131733655929565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,4096,0.019592533508936562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,5120,0.016155733664830526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,3584,0.010160000125567118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,3584,0.018644267320632936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,4096,0.013643733660380044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,3072,0.009177600344022114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,3072,0.016800000270207723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,3584,0.013128532965977987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,2560,0.008039466540018718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,2560,0.016876800855000814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,3072,0.011597866813341778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,2048,0.006856533388296763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,2048,0.015083733201026916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,2560,0.011086933811505635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,1536,0.005551999807357788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,1536,0.014674133062362671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,2048,0.009958400328954061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,1024,0.0034272000193595886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,1024,0.01456106702486674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,1024,0.008338133494059246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,768,0.0030293333033720653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,768,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,1536,0.009149866302808125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,512,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,512,0.01448853313922882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,512,0.007918933530648549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,256,0.002595199892918269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,768,0.007513600091139476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,256,0.01360640029112498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,128,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,128,0.013347199559211731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,128,0.007090133428573608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,64,0.002204799900452296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,64,0.013924266894658408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,7168,32,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,7168,32,0.013781332969665527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,256,0.007057066758473713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,65536,0.12041066487630207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,65536,0.0815936009089152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,16384,0.0378165324529012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,16384,0.03247146606445313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,8192,65536,0.12558826605478923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,12288,0.027427200476328534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,12288,0.03134933312733968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,16384,0.035889065265655516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,10240,0.02271359960238139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,10240,0.0270197331905365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,12288,0.028445865710576373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,8192,0.019036799669265747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,8192,0.023704532782236734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,8192,0.021505065759023032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,7168,0.0172106663386027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,7168,0.023121066888173423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,10240,0.025192532936731977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,6144,0.01548479994138082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,6144,0.02035413384437561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,7168,0.019802665710449217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,5120,0.013682132959365845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,5120,0.02050559918085734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,6144,0.017749333381652833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,4096,0.010048000017801921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,4096,0.01999893387158712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,4096,0.013340800007184347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,3584,0.009318400422732036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,3584,0.019432532787322997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,5120,0.015752533078193666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,3072,0.00839573343594869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,3072,0.016619732975959776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,3584,0.012449066837628682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,2560,0.007495466868082683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,2560,0.014987732966740927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,3072,0.011733333269755047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,2048,0.006533333162466686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,2048,0.0163263996442159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,2048,0.010054399569829304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,1536,0.004678399860858917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,1536,0.014666666587193808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,2560,0.010825600226720173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,1024,0.0034101332227389016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,1024,0.014614400267601014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,1024,0.007950933277606964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,768,0.003005866706371307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,768,0.01452906628449758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,1536,0.009127466877301534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,512,0.00264533335963885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,512,0.014598400394121806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,7168,65536,0.12532693545023602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,256,0.0025386666258176167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,256,0.013460266590118408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,256,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,768,0.007538133362929027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,128,0.0022272000710169475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,128,0.013800533612569174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,64,0.0021930667261282604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,512,0.007540266712506611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,64,0.013459199666976928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,6144,32,0.0022282667458057405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,6144,32,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,128,0.0066655998428662615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,65536,0.10158826510111492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,65536,0.07254719734191895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,16384,0.03302293419837952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,16384,0.03161279956499736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,12288,0.023948800563812257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,12288,0.027270400524139406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,12288,0.028471465905507403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,10240,0.020096000035603842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,10240,0.025123200813929242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,16384,0.03551040093104045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,8192,0.017400532960891724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,8192,0.02072426676750183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,10240,0.025224532683690386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,7168,0.015416533748308817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,7168,0.020691200097401937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,8192,0.02113173405329386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,6144,0.013713066776593527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,6144,0.01939093271891276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,7168,0.019489065806070963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,5120,0.012089600165685017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,5120,0.018942934274673463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,6144,0.017334399620691936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,4096,0.009221333265304565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,4096,0.019108267625172932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,5120,0.01618773341178894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,3584,0.00886826713879903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,3584,0.016506666938463845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,4096,0.01372053325176239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,3072,0.007769600053628286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,3072,0.01470186710357666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,3584,0.012947199741999307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,2560,0.00727040022611618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,2560,0.016477866967519125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,3072,0.01162453293800354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,2048,0.005494399865468343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,2048,0.015706666310628257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,2048,0.009956266482671101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,1536,0.003818666686614355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,2560,0.011070932944615681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,1536,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,1024,0.0034954667091369627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,1024,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,1024,0.007923200229803721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,768,0.003014400104681651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,768,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,1536,0.009269332885742188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,512,0.002739199995994568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,512,0.014597333470980325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,768,0.007909333209196727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,256,0.0021727999051411946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,6144,65536,0.1257695992787679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,256,0.01404906709988912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,128,0.002233600119749705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,128,0.013454932967821756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,128,0.0067071999112765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,64,0.0021664001047611236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,512,0.00754559983809789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,64,0.013867732882499696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,5120,32,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,256,0.0075530668099721265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,5120,32,0.013080533345540366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,65536,0.08250239690144857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,65536,0.05965439875920614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,16384,0.025912533203760784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,16384,0.028358399868011475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,12288,0.01950719952583313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,12288,0.024489599466323852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,12288,0.028908799091974895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,10240,0.016622933745384216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,10240,0.023774933815002442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,16384,0.03581440051396688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,8192,0.013677866260210673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,8192,0.021310933430989585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,10240,0.025177599986394246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,7168,0.012601600090662638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,7168,0.025500800212224322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,8192,0.02108373244603475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,6144,0.01163093348344167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,6144,0.02069546580314636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,7168,0.01949440042177836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,5120,0.010496000448862713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,5120,0.020780799786249797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,6144,0.017433599630991618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,4096,0.009284266829490661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,4096,0.019899733861287437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,5120,0.015749333302179973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,3584,0.0115146666765213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,3584,0.01800640026728312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,4096,0.01325653294722239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,3072,0.009223467111587525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,3072,0.016557866334915163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,3584,0.012874666849772134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,2560,0.008361599842707316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,2560,0.0165802667538325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,3072,0.011689600348472596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,2048,0.006694399813810985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,2048,0.016523733735084534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,2560,0.011186133821805317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,1536,0.0059125334024429325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,1536,0.015660799543062844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,1536,0.009562666217486065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,1024,0.0046528001626332605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,1024,0.015433599551518759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,2048,0.009970133503278095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,768,0.0038133333126703895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,768,0.01453013320763906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,768,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,512,0.0034101332227389016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,512,0.014098133643468222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,1024,0.008423466483751934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,256,0.002629333237806956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,256,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,256,0.007123200098673503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,128,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,128,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,512,0.007157333195209503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,64,0.002169599880774816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,64,0.013969066739082336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,4096,32,0.0021557333568731946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,128,0.006718933085600535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,4096,32,0.013174399733543396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,65536,0.07147839864095053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,65536,0.05633813142776489
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,16384,0.02269866665204366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,16384,0.02646399935086568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,5120,65536,0.12522880236307782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,12288,0.01684373418490092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,12288,0.025665066639582318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,12288,0.028540800015131634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,10240,0.014893866578737893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,10240,0.02471253275871277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,16384,0.035588268438975015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,8192,0.012596266468365988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,8192,0.02147093415260315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,10240,0.025123200813929242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,7168,0.011904000242551168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,7168,0.022252800067265828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,8192,0.020716800292332967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,6144,0.010834133625030518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,6144,0.02177919944127401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,7168,0.019053866465886436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,5120,0.009783466657002766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,5120,0.019099734226862588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,6144,0.016976000865300496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,4096,0.008717866738637288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,4096,0.017485866943995156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,5120,0.015492266416549683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,3584,0.010477866729100544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,3584,0.018067200978597008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,4096,0.013276799519856771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,3072,0.009161600470542907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,3072,0.017571200927098594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,3584,0.01250986655553182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,2560,0.008040533463160197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,3072,0.01162559986114502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,2560,0.016902399063110352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,2048,0.007113599777221679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,2048,0.016523733735084534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,4096,65536,0.1249557336171468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,1536,0.005524266759554545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,1536,0.014590932925542196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,2048,0.010126933455467224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,1024,0.004689066608746847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,1024,0.01495146652062734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,2560,0.010894933342933654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,768,0.003958400090535482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,1536,0.009205333391825358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,768,0.014542933305104574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,512,0.0035306667288144433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,512,0.013781332969665527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,512,0.0075434664885203045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,256,0.0022677332162857056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,768,0.0075530668099721265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,256,0.014406399925549826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,128,0.002209066599607468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,128,0.013452800114949546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,1024,0.008428800106048583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,64,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,64,0.014435199896494546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,256,0.0070602665344874065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,128,0.006786133348941803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3584,32,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3584,32,0.013658666610717773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,65536,0.06228266557057699
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,16384,0.019432532787322997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,65536,0.051164801915486655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,16384,0.025590399901072185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,12288,0.01498133341471354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,12288,0.023175466060638427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,12288,0.02831573287645976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,10240,0.013657599687576294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,10240,0.020863999923070274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,16384,0.03580693403879802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,8192,0.011718400319417318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,8192,0.020833067099253335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,10240,0.024714666604995727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,7168,0.010814932982126872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,7168,0.02065599958101908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,8192,0.02081813414891561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,6144,0.010050132870674133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,6144,0.020755199591318767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,7168,0.01923946738243103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,5120,0.009198932846387228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,5120,0.019099734226862588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,6144,0.017409066359202065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,4096,0.00823466678460439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,4096,0.01857066750526428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,5120,0.015738667050997416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,3584,0.009264000256856282
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,3584,0.016535466909408568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,4096,0.013291733463605246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,3072,0.008583466211954754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,3072,0.017595734198888144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,3584,0.012475732962290447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,2560,0.00668693333864212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,2560,0.016668800512949625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,3072,0.011636267105738323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,2048,0.006340266764163971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,2048,0.017003732919692992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3584,65536,0.1252832015355428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,1536,0.005456000069777171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,1536,0.01609599987665812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,2560,0.011196800072987874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,1024,0.004273066421349844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,1024,0.015364266435305276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,2048,0.010342400272687275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,768,0.003869866579771042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,768,0.015147733688354491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,1536,0.009222400188446046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,512,0.003398400048414866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,512,0.01446399986743927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,768,0.007881600161393483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,256,0.002918400118748347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,256,0.014435199896494546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,1024,0.008317866424719492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,128,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,128,0.013287466764450074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,512,0.0075093333919843035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,64,0.0021717332303524016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,64,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,256,0.0071381335457166035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,3072,32,0.0021429332594076794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,128,0.006697600086530049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,3072,32,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,65536,0.053208533922831217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,16384,0.017824000120162962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,65536,0.04583679835001628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,16384,0.022983467578887938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,12288,0.013229866822560629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,12288,0.021729065974553426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,12288,0.028474666674931842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,10240,0.012180266777674358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,10240,0.02067413330078125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,16384,0.035819733142852785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,8192,0.010426666339238484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,8192,0.020578134059906005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,10240,0.024910932779312132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,7168,0.009776000181833904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,7168,0.018784000476201376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,8192,0.0206986665725708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,6144,0.008824533224105835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,6144,0.018623999754587808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,7168,0.019160532951354982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,5120,0.008066133161385854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,5120,0.01872426668802897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,6144,0.017027199268341064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,4096,0.007114666700363159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,4096,0.01700586676597595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,5120,0.015777066349983215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,3584,0.006306133170922597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,3584,0.017627733945846557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,4096,0.013637333114941915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,3072,0.007177599767843883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,3072,0.01657600005467733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,3584,0.0129120002190272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,2560,0.005914666752020518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,2560,0.01655466655890147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,3072,0.011640533804893494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,2048,0.006286933521429698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,2048,0.016459733247756958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,2560,0.011202133695284526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,1536,0.005411200225353241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,1536,0.015640532970428465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,2048,0.009965866804122925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,1024,0.0045738667249679565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,1024,0.014686933159828186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,1024,0.008348799745241801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,768,0.003833599885304769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,768,0.014522666732470194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,1536,0.009248000383377076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,3072,65536,0.1249951998392741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,512,0.0031136001149813337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,512,0.014565333724021912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,256,0.0026975999275843303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,256,0.014519466956456503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,768,0.0075445334116617845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,128,0.0026133333643277483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,256,0.007147733370463054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,512,0.0075914666056633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,128,0.01328000028928121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,64,0.002203733225663503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,64,0.013959466417630514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2560,32,0.002198400100072225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,128,0.006783999999364217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2560,32,0.013116799791653953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,65536,0.04347199996312459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,65536,0.04102720022201538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,16384,0.015384533007939658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,16384,0.022712532679239908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,12288,0.011638399958610535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,12288,0.021176532904307047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,12288,0.027781333525975543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,10240,0.010135466853777569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,10240,0.02044266661008199
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,16384,0.035145600636800126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,8192,0.009465600053469341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,8192,0.02074880003929138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,10240,0.024567466974258424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,8192,0.020702934265136717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,7168,0.00906880001227061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,7168,0.020862932999928793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,6144,0.0079925333460172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,6144,0.01723519961039225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,6144,0.01734506686528524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,5120,0.006733866532643636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,5120,0.01653866668542226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,7168,0.019218132893244425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,4096,0.008575999736785888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,4096,0.018785067399342856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,5120,0.015698132912317912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,3584,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,3584,0.017272533973058064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,4096,0.013762133320172629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,3072,0.005997866888840993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,3072,0.016682666540145875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,3584,0.012414933244387309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,2560,0.005067733426888784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,2560,0.016505600015322367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,3072,0.012176000078519185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,2048,0.006467199822266896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,2048,0.016636799772580466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2560,65536,0.1253717343012492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,1536,0.00561706672112147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,2560,0.010789333780606588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,1536,0.01593386630217234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,1024,0.0044064000248909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,1024,0.015034666657447815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,2048,0.010009599725405376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,768,0.0038442666331926978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,768,0.01471466620763143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,1024,0.008327466746171314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,512,0.003436800092458725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,768,0.008002133170763651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,1536,0.00874773363272349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,256,0.002631466587384542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,512,0.014662399888038635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,256,0.014291200041770934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,128,0.0023103999594847363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,128,0.013489066561063131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,256,0.007206400235493978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,128,0.006651733318964641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,64,0.002299733211596807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,2048,32,0.002238933245340983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,64,0.013080533345540366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,512,0.00732479989528656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,65536,0.03380053440729777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,2048,32,0.014551466703414917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,65536,0.03515413204828898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,16384,0.012077866991360982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,16384,0.024808533986409507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,12288,0.01492693324883779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,12288,0.023561600844065347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,12288,0.02772480050722758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,10240,0.01292693316936493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,10240,0.022737065951029457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,16384,0.03472213347752889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,8192,0.011256532867749532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,8192,0.020508799950281778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,10240,0.02440320054690043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,7168,0.010831999778747558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,7168,0.02059626579284668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,8192,0.020336000124613444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,6144,0.009686400492986042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,6144,0.018397865692774455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,7168,0.018694400787353516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,5120,0.008473599950472515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,5120,0.01665493349234263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,6144,0.01703146696090698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,4096,0.007221333185831706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,4096,0.017452800273895265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,5120,0.015681067109107973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,3584,0.006820266445477803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,3584,0.018133334318796792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,4096,0.013640532890955607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,3072,0.005562666555245717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,3072,0.016619732975959776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,3584,0.012446932991345723
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,3072,0.011588266491889954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,2560,0.005118933320045471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,2560,0.016662399967511496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,2048,0.005364266534646353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,2048,0.014636799693107605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,2560,0.010787199934323628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,1536,0.005067733426888784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,1536,0.014626133441925048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,1536,0.009250133236249288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,1024,0.00420906643072764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,1024,0.014800000190734863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,2048,0.009961600104967754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,768,0.0034175999462604523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,768,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,2048,65536,0.12153279781341553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,512,0.0030559999247392017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,512,0.014589866995811463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,768,0.007501866420110066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,256,0.002619733413060506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,1024,0.008333866794904072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,256,0.014387200276056925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,128,0.0022912000616391497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,512,0.007560533285140991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,128,0.014120533068974813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,64,0.0021397332350413007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,64,0.014060800274213156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,256,0.00705813318490982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,128,0.00673280010620753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1536,32,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1536,32,0.01444906691710154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,65536,0.024222934246063234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,65536,0.028894933064778645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,16384,0.01365546683470408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,16384,0.020693333943684895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,12288,0.010757333040237427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,12288,0.01958613395690918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,12288,0.027609600623448687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,10240,0.00939520001411438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,10240,0.01935466726620992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,16384,0.03548266490300496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,8192,0.008771199981371562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,8192,0.019671465953191122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,10240,0.02448106606801351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,7168,0.007574399809042613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,7168,0.019887999693552653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,8192,0.021157334248224892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,6144,0.006749866902828217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,6144,0.01773973306020101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,7168,0.019011199474334717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,5120,0.006313600142796834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,5120,0.01650773286819458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,6144,0.017790933450063072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,4096,0.006222933530807495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,4096,0.01798506577809652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,5120,0.015723733107248943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,3584,0.005492266515890757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,3584,0.018029866615931193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,4096,0.013689600427945457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,3072,0.005443199972311656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,3072,0.017517866690953572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,3584,0.012492799758911132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,2560,0.005048533280690512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,2560,0.0164874662955602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,3072,0.012014933427174886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,2048,0.006202666461467743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,2048,0.015492266416549683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,2560,0.010762666662534077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,1536,0.005030400057633718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,1536,0.014450132846832275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,2048,0.010375466942787171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,1024,0.0038399999340375268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,1024,0.014441600441932679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,1536,0.00921066701412201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,768,0.0034261333445707956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,768,0.0145578662554423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,1024,0.008306133250395458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,512,0.0030400000512599947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,512,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,768,0.007602133353551229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,256,0.0025610665480295816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,256,0.01416213313738505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,512,0.00790826678276062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,128,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,256,0.007089066505432129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,128,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,64,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,64,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,128,0.0063360000650088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,1024,32,0.0021845333278179167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,65536,0.020754132668177286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,1024,32,0.013430399696032205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,65536,0.03341439962387085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,16384,0.011449600259462994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,16384,0.02098133365313212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1536,65536,0.12200106779734295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,12288,0.009655466675758362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,12288,0.020472532510757445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,12288,0.02760533293088277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,10240,0.007884799937407176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,10240,0.01832853356997172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,16384,0.03504746754964193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,8192,0.007274666428565979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,8192,0.01872319976488749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,10240,0.02487893303235372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,7168,0.006863999863465626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,7168,0.018692266941070557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,8192,0.020639999707539877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,6144,0.006758399804433187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,6144,0.018127999703089395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,6144,0.01733760039011637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,5120,0.005527466535568237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,5120,0.01687893271446228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,7168,0.019504000743230186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,4096,0.006009600063165029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,4096,0.01748159925142924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,5120,0.015290666619936624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,3584,0.00582826683918635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,4096,0.013724799950917563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,3584,0.017067732413609822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,1024,65536,0.12079573472340902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,3072,0.005448533097902933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,3072,0.016746666034062704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,2560,0.005604266623655955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,2560,0.016858667135238647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,3072,0.011597866813341778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,2048,0.004228266576925913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,2048,0.015097600221633912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,3584,0.012995200355847678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,1536,0.004931200047334035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,2560,0.010824533303578694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,1536,0.014551466703414917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,1024,0.003912533322970072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,1024,0.01471573313077291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,1536,0.009248000383377076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,768,0.0035242666800816857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,1024,0.008425600330034892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,2048,0.009546666344006857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,512,0.0029866665601730345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,768,0.014707199732462563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,512,0.014443733294804893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,256,0.0025802666942278544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,256,0.013402666648228964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,768,0.008470400174458822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,512,0.0075445334116617845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,128,0.0022944000860055285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,256,0.006900266806284587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,128,0.014501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,64,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,768,32,0.002164266755183538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,128,0.0067210664351781205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,64,0.01444906691710154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,65536,0.02765866716702779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,768,32,0.013430399696032205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,16384,0.007915733257929485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,65536,0.0291701336701711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,16384,0.017591466506322227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,12288,0.0075082664688428235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,12288,0.018576000134150186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,12288,0.028025599320729573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,10240,0.006705066561698914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,10240,0.01815999945004781
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,16384,0.03477866649627685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,8192,0.00561706672112147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,8192,0.01882666746775309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,10240,0.024318933486938477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,7168,0.005478399991989136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,7168,0.01914880077044169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,8192,0.02081706722577413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,6144,0.0060138667623202005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,6144,0.01853760083516439
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,7168,0.018974934021631876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,5120,0.005470933516820272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,5120,0.016581333676973977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,6144,0.017490132649739584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,4096,0.005970133344332377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,4096,0.017824000120162962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,5120,0.015726932883262636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,3584,0.005480533341566721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,3584,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,4096,0.013312000036239623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,3072,0.005542399982611338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,3584,0.01241600016752879
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,3072,0.01618559956550598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,2560,0.005123200019200643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,768,65536,0.12037119865417481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,2560,0.016639999548594155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,2048,0.00479360024134318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,2048,0.016617600123087564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,3072,0.01165120005607605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,1536,0.003790933390458425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,1536,0.014475733041763306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,2560,0.010859733819961548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,1024,0.003124266614516576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,2048,0.009577600161234538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,1024,0.014553599556287131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,768,0.00310506671667099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,768,0.014648532867431641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,1536,0.008726400136947633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,1024,0.008371200164159138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,512,0.0026378666361172995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,512,0.014575999975204468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,768,0.007559466858704884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,256,0.00271573339899381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,256,0.01384106675783793
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,512,0.007063466807206471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,128,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,256,0.006693333387374878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,128,0.014421332875887552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,64,0.0021386665602525076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,64,0.014203733205795288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,128,0.0063178668419520065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,512,32,0.0022122666239738466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,512,32,0.014082133769989014
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,65536,0.016498133540153503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,65536,0.023717333873113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,16384,0.00553173323472341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,16384,0.01879040002822876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,12288,0.005238399902979533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,12288,0.018976000944773357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,16384,0.03508373498916626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,10240,0.005588266750176748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,10240,0.018905599912007652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,12288,0.027560534079869588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,8192,0.005410133302211762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,8192,0.016769067446390788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,10240,0.02442453304926554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,7168,0.005096533397833506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,7168,0.018631466229756675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,8192,0.02062186598777771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,6144,0.005016533533732096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,6144,0.016556800405184428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,7168,0.019011199474334717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,5120,0.004722133278846741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,6144,0.0173418660958608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,5120,0.01665600041548411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,4096,0.004370133578777313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,4096,0.016621866822242738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,5120,0.015709867080052696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,3584,0.005016533533732096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,3584,0.016886399189631144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,4096,0.013389866550763449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,3072,0.004704000055789947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,3072,0.015127467115720114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,3584,0.012417067090670269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,2560,0.004956800242265066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,512,65536,0.12061546643575032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,2560,0.014617600043614707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,2048,0.0046015997727712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,3072,0.011705600221951802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,2048,0.014787200093269347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,1536,0.0038015998899936674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,1536,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,2048,0.00999786655108134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,2560,0.011320533355077107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,1024,0.003058133274316788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,1536,0.009129599730173747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,1024,0.01454080045223236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,768,0.002995199958483378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,768,0.014532267053922018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,1024,0.00832426647345225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,512,0.0026687999566396077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,768,0.0075647999842961625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,512,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,256,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,256,0.013111467162768045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,512,0.007549866537253062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,128,0.002199466774861018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,256,0.007064533233642578
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,128,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,64,0.0021802666286627453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,256,32,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,128,0.006362666686375936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,64,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,65536,0.00920746624469757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,256,32,0.01365013321240743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,16384,0.005496533215045929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,65536,0.02071466644605001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,16384,0.019011199474334717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,12288,0.005055999755859375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,12288,0.01759999990463257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,12288,0.027998934189478557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,10240,0.005465599894523621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,16384,0.03504000107447307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,256,65536,0.11960000197092693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,10240,0.018849066893259683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,8192,0.00547626664241155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,8192,0.016578132907549538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,7168,0.004700799783070883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,7168,0.01768746574719747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,10240,0.024829866488774617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,8192,0.020733867088953653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,6144,0.005066666503747304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,7168,0.019028266270955406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,6144,0.016764799753824867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,5120,0.0042922665675481165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,5120,0.01702293356259664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,4096,0.004309333364168803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,4096,0.01812160015106201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,6144,0.017366399367650352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,5120,0.015380266308784484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,3584,0.004709333181381226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,3584,0.015800533692042033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,4096,0.013670399785041809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,3072,0.004677333434422811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,3072,0.015844266613324484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,3584,0.012900267044703165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,2560,0.004689066608746847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,2560,0.014495999614397685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,3072,0.011654399832089742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,2048,0.004670933385690053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,2048,0.015576533476511636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,2560,0.011204266548156738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,1536,0.0035989334185918174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,1536,0.014568533500035605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,2048,0.00959999958674113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,1024,0.003049599876006444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,1024,0.01502826710542043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,1536,0.009546666344006857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,768,0.002977066735426585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,1024,0.007950933277606964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,768,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,512,0.002666666607062022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,768,0.007886933286984761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,512,0.013911466797192892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,256,0.002349866678317388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,512,0.007158400118350982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,256,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,65536,0.11941013336181641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,128,0.0022261333962281544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,256,0.006300800045331319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,128,0.012920533617337545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,64,0.0021397332350413007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,128,32,0.0022122666239738466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,8,128,128,0.00653546651204427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,64,0.013041067123413085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,65536,0.007150933146476746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,128,32,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,16384,0.005512533088525137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,65536,0.020536533991495767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,12288,0.005073066552480062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,16384,0.017875200510025023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,10240,0.004715733230113983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,12288,0.018521600961685182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,10240,0.01773866613705953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,8192,0.005545599758625031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,8192,0.017707733313242595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,7168,0.005133866767088572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,5120,0.0042303999265035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,6144,0.0052138666311899815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,7168,0.017770665884017944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,6144,0.017887999614079796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,4096,0.004343466460704803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,3584,0.004997333387533823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,5120,0.01653333306312561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,4096,0.016721065839131674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,3072,0.0046528001626332605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,2560,0.004642133414745331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,3072,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,2048,0.004016000032424927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,2560,0.014595199624697366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,3584,0.01586560010910034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,1536,0.003751466671625773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,2048,0.014617600043614707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,1024,0.0030741333961486817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,768,0.002625066787004471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,1536,0.014571733276049294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,1024,0.014709333578745524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,512,0.002573866645495097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,256,0.0022133332987626392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,128,0.0022154666483402254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,512,0.01388159990310669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,768,0.014600533246994018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,256,0.012757333119710288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,64,0.0021727999051411946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,64,32,0.0018154667069514591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,128,0.014391466975212097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,65536,0.00666133314371109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,32,0.014537599682807923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,16384,0.005022933085759481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,64,64,0.012817066907882691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,65536,0.02031146685282389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,12288,0.0050805335243542995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,16384,0.016563199957211814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,10240,0.00513919989267985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,8192,0.00421973317861557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,12288,0.017190400759379068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,10240,0.016590933005015053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,7168,0.004582400123278299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,8192,0.01665279964605967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,6144,0.004796800017356872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,5120,0.004283733169237773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,6144,0.016528000434239708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,4096,0.004192000130812327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,7168,0.017806933323542277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,5120,0.01738986571629842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,3584,0.004259199897448222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,3072,0.004628266890843709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,4096,0.017550933361053466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,2560,0.004645333190759023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,3072,0.01604373355706533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,3584,0.01568000018596649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,2048,0.0046634669105211895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,1536,0.003432533393303553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,2560,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,1024,0.0030005333324273427
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,2048,0.015174399813016257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,1536,0.014507733782132468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,768,0.0026869334280490874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,1024,0.014434132973353067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,512,0.0022485333184401194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,768,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,256,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,512,0.01318186620871226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,128,0.0022229333718617756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,256,0.01381013294061025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,64,0.0020288000504175823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,128,0.013061333696047464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,8,32,32,0.0018314667046070098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,64,0.013399466872215271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,8,32,32,0.012658133109410604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,16384,0.3203488032023112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,16384,0.17817494074503581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,12288,0.23685439427693686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,12288,0.1396074612935384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,10240,0.2009162743886312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,10240,0.11951253414154053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,8192,0.16236693064371746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,8192,0.093776003519694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,8192,0.06634453137715658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,7168,0.142521603902181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,7168,0.08323520024617513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,10240,0.08074560165405273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,6144,0.1257205327351888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,6144,0.07318933010101318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,12288,0.10357653299967448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,5120,0.10502933661142985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,5120,0.06371093193689982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,6144,0.053491199016571046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,4096,0.08005546728769938
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,4096,0.054161067803700766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,7168,0.06131519873936971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,3584,0.06955093542734782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,3584,0.0494762659072876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,5120,0.04554239908854167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,3072,0.060908798376719156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,3072,0.04490240017573039
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,16384,0.1364138603210449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,2560,0.05144960085550944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,2560,0.04165759881337484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,3584,0.034622931480407716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,2048,0.04113279978434245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,2048,0.035394132137298584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,4096,0.038644266128540036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,1536,0.03204266627629598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,1536,0.029806933800379437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,3072,0.03227413296699524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,1024,0.022156800826390585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,1024,0.024730666478474935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,2560,0.027720532814661664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,768,0.017590399583180746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,768,0.022471467653910317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,2048,0.022759467363357544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,512,0.012636799613634744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,512,0.02027626633644104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,1536,0.019735467433929444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,256,0.007512533167997997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,256,0.016635732849438985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,256,0.009219200412432352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,128,0.0046528001626332605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,128,0.016535466909408568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,512,0.010855467120806376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,64,0.004307200014591217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,64,0.01648319959640503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,65536,32,0.004332800209522247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,65536,32,0.01646719972292582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,65536,0.3233855883280436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,65536,0.17844586372375487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,128,0.008340266346931458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,16384,0.09262932936350504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,16384,0.06134080092112223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,768,0.013758933544158936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,12288,0.0664031982421875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,12288,0.04319680134455363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,65536,1024,0.015735466281572977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,10240,0.056713600953420006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,10240,0.03927573362986247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,10240,0.03441280126571655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,8192,0.047525334358215335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,8192,0.0343018651008606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,12288,0.04069120089213053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,7168,0.035596799850463864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,7168,0.032943999767303465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,16384,0.04756373167037964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,6144,0.03138239979743958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,6144,0.029180800914764403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,8192,0.028460800647735596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,5120,0.026666667064030963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,5120,0.026844799518585205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,7168,0.025196800629297893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,4096,0.021862399578094483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,4096,0.02416320045789083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,5120,0.020101332664489747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,3584,0.019132800896962485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,3584,0.022721066077550253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,6144,0.022833067178726196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,3072,0.01710933248202006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,3072,0.02256960074106852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,4096,0.01723946730295817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,2560,0.014574933052062988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,2560,0.020797866582870483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,3072,0.014913066228230795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,2048,0.01237440009911855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,2048,0.018874667088190713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,2560,0.012826666235923767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,1536,0.00960640013217926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,1536,0.017812265952428182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,3584,0.01622719963391622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,1024,0.007322666545708974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,1024,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,1024,0.008739200234413148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,768,0.005924266576766968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,768,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,1536,0.010523733496665955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,512,0.003555200000603994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,512,0.014607999722162882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,768,0.008310399949550629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,256,0.002639999985694885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,256,0.01416106621424357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,512,0.007976533472537994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,128,0.0026634665826956432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,128,0.014284800489743552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,256,0.007117866476376851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,64,0.002196266750494639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,2048,0.012061867117881774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,64,0.013025066256523133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,16384,32,0.0023050665855407714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,16384,32,0.01376426617304484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,128,0.0070816000302632645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,65536,0.23408427238464355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,65536,0.13997546831766766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,16384,0.06953919728597005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,16384,0.0494922677675883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,12288,0.04943893353144328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,12288,0.0408021330833435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,12288,0.030010666449864703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,10240,0.04243626594543457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,10240,0.03671679894129436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,16384,0.038389333089192706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,8192,0.03626559972763062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,8192,0.03294720053672791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,10240,0.025963733593622845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,7168,0.03139839967091878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,7168,0.031176533301671343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,8192,0.022281599044799805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,6144,0.02758293350537618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,6144,0.028806400299072266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,7168,0.020253866910934448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,5120,0.023561600844065347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,5120,0.0268010675907135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,6144,0.018720000982284546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,4096,0.0201365331808726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,4096,0.025166932741800947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,5120,0.01696746746699015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,3584,0.01755519906679789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,3584,0.020602667331695558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,4096,0.01453013320763906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,3072,0.015869866808255514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,3072,0.02048426667849223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,3584,0.013825066884358726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,2560,0.013705600301424661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,2560,0.01904746691385905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,3072,0.012923733393351237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,2048,0.012570666273434958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,2048,0.01835839947064718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,2560,0.011248000462849935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,1536,0.008184533317883809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,1536,0.015361066659291586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,1536,0.0092330664396286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,1024,0.00629013329744339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,1024,0.014542933305104574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,2048,0.01051626702149709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,768,0.0043722664316495265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,768,0.014632532993952433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,768,0.007977599898974102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,512,0.003491200009981791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,512,0.014678399761517844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,1024,0.008360532919565837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,256,0.00266239990790685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,256,0.013031466801961263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,256,0.0070933332045873005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,512,0.007669333120187123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,128,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,128,0.014512000481287637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,64,0.0022805333137512207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,64,0.014537599682807923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,12288,32,0.0023221333821614584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,12288,32,0.014588800072669984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,65536,0.1956000010172526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,128,0.006729599833488464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,65536,0.12354453404744466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,16384,0.060363733768463136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,16384,0.043858134746551515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,16384,0.037062398592631024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,12288,0.042182401816050215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,12288,0.037088000774383546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,12288,0.029121067126591998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,10240,0.03636266787846883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,10240,0.03333866596221924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,16384,65536,0.16528213818868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,8192,0.029072000583012896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,8192,0.030726399024327593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,10240,0.025455999374389648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,7168,0.02709226608276367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,7168,0.029179733991622925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,8192,0.021894399325052896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,6144,0.023661865790685018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,6144,0.026825600862503053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,7168,0.02007360061009725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,5120,0.020503467321395873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,5120,0.026081067323684693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,12288,65536,0.135207462310791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,4096,0.016884267330169678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,4096,0.02074986696243286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,6144,0.0184608002503713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,3584,0.015778133273124696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,3584,0.021284266312917074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,5120,0.016657066345214844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,3072,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,3072,0.01973973313967387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,4096,0.014484266440073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,2560,0.012567466497421265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,2560,0.018567466735839845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,3584,0.013482667009035745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,2048,0.009662933150927226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,2048,0.01874026656150818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,2560,0.010873599847157796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,1536,0.007693866888682048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,1536,0.016518400112787882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,2048,0.009964799880981446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,1024,0.005854933460553487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,1024,0.016577066977818807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,3072,0.012454400459925335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,768,0.0046741331617037455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,768,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,768,0.007997866471608479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,512,0.0042357335488001505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,512,0.014497066537539164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,1024,0.008400000135103862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,256,0.0026069333155949908
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,256,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,512,0.007516799867153168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,128,0.002205866575241089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,256,0.007125333448251088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,128,0.013613866766293845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,64,0.0021503999829292296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,64,0.013779200116793313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,10240,32,0.0021770666042963665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,10240,32,0.013186132907867432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,1536,0.009204266468683879
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,65536,0.15817707379659016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,65536,0.09773546854654948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,16384,0.048596266905466715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,128,0.006782933572928111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,16384,0.03546239932378133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,12288,0.034644265969594315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,12288,0.030405332644780476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,12288,0.028465066353480024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,10240,0.02980053424835205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,10240,0.028037333488464357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,16384,0.03625173171361287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,8192,0.022166399161020915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,8192,0.02569493254025777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,10240,0.02515946626663208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,7168,0.019489065806070963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,7168,0.022792534033457438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,8192,0.021709867318471274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,6144,0.017473065853118898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,6144,0.022707200050354003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,7168,0.020071466763814293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,5120,0.014968533317248026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,5120,0.02116159995396932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,10240,65536,0.12752319971720377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,4096,0.012299733360608418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,4096,0.02015786568323771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,6144,0.018132267395655315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,3584,0.011202133695284526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,3584,0.018994132677714028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,5120,0.016165332992871602
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,3072,0.0100000003973643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,3072,0.018609066804250084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,4096,0.014104533195495605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,2560,0.00876693328221639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,2560,0.01694933374722799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,3584,0.013041067123413085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,2048,0.007729066908359528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,2048,0.01669973333676656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,3072,0.012196266651153564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,1536,0.006355200211207073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,1536,0.01458346645037333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,2048,0.010037333766619364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,2560,0.011248000462849935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,1024,0.0038463999827702843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,1024,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,768,0.00393599991997083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,768,0.014922666549682616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,1024,0.008409600456555684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,512,0.0029898665845394133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,512,0.01411946713924408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,1536,0.009282132983207703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,256,0.0027253332237402597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,256,0.01402239998181661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,768,0.00796693315108617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,128,0.0022986667851607008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,128,0.014225066701571146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,128,0.006853333115577698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,64,0.002271999915440877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,512,0.007518933216730754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,64,0.014525866508483887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,8192,32,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,256,0.00714026689529419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,8192,32,0.014206932981808982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,65536,0.1384928067525228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,65536,0.08901333014170329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,16384,0.04011413256327311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,16384,0.03521173397699992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,12288,0.030429865916570025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,12288,0.02748799920082092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,12288,0.02865920066833496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,10240,0.02632213234901428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,10240,0.025326933463414508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,16384,0.035995733737945554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,8192,0.021806933482487998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,8192,0.024406399329503378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,10240,0.025387734174728394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,7168,0.019789866606394448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,7168,0.022680532932281495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,8192,0.021571199099222817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,6144,0.015759999553362526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,6144,0.020719999074935914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,7168,0.01988160014152527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,5120,0.01383680005868276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,5120,0.020371200640996297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,6144,0.018282665808995565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,4096,0.01127893328666687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,4096,0.018731733163197838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,5120,0.016376533110936484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,3584,0.010328533252080281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,3584,0.018600533405939736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,4096,0.013921067118644714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,3072,0.009160533547401428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,3072,0.017199999094009398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,3584,0.01285866697629293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,2560,0.008434133728345235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,2560,0.01691840092341105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,3072,0.01162559986114502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,2048,0.007134933272997539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,2048,0.015107199549674988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,2560,0.010915199915568035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,1536,0.0056426664193471275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,1536,0.014632532993952433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,2048,0.009975467125574749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,1024,0.003815466662247976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,1024,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,1536,0.009286399682362874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,768,0.0033952000240484873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,768,0.014548266927401224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,1024,0.007968000074227651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,512,0.0029557332396507262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,512,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,512,0.007496533294518788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,256,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,256,0.013409066200256347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,768,0.00798933357000351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,128,0.002274133265018463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,128,0.013541332880655923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,128,0.006840533514817555
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,64,0.0022815999885400133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,256,0.0070783997575442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,7168,32,0.0021557333568731946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,64,0.01317759950955709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,7168,32,0.012917332847913108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,65536,0.11981653372446696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,65536,0.0824949343999227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,16384,0.037367467085520426
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,16384,0.03282986680666606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,8192,65536,0.12605013052622477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,12288,0.028245333830515546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,12288,0.02847893238067627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,16384,0.035844266414642334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,10240,0.02360960046450297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,10240,0.026581333080927534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,12288,0.02892373402913411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,8192,0.019692800442377725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,8192,0.02276159922281901
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,10240,0.025196800629297893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,7168,0.01744746764500936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,7168,0.02262506683667501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,8192,0.021565866470336915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,6144,0.01568000018596649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,6144,0.020713599522908528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,7168,0.019836799303690592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,5120,0.013749333222707114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,5120,0.01899306575457255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,6144,0.018216532468795777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,4096,0.010634666681289673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,4096,0.01866133411725362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,5120,0.01583466629187266
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,3584,0.009433600306510925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,3584,0.01741973360379537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,4096,0.013820800185203552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,3072,0.008612266182899475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,3072,0.017065600554148356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,3584,0.012519466876983642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,2560,0.007566933333873749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,2560,0.01654293338457743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,3072,0.012081066767374676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,2048,0.0068010667959849044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,2048,0.01458560029665629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,7168,65536,0.1255413293838501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,1536,0.004771199822425842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,1536,0.0147189329067866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,1536,0.009342933694521587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,1024,0.003934933245182038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,1024,0.01481066644191742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,2048,0.010531199971834819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,768,0.0035242666800816857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,2560,0.010830932855606079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,768,0.014573867122332254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,512,0.003083733220895131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,512,0.014173866311709086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,512,0.007607466479142506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,256,0.0026890667776266735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,256,0.013844266533851624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,1024,0.00839573343594869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,128,0.0022624000906944276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,768,0.007499733567237854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,128,0.014636799693107605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,64,0.0022453332940737406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,64,0.014524799585342408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,6144,32,0.0021749332547187803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,128,0.006663466493288676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,6144,32,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,65536,0.10145066579182942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,65536,0.07187519868214926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,256,0.007112533350785573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,16384,0.032935466368993124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,16384,0.031122134129206343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,12288,0.024623999993006386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,12288,0.026649600267410277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,12288,0.02902719974517822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,10240,0.02063039938608805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,10240,0.024684800704320272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,16384,0.035802666346232095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,8192,0.01744426687558492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,8192,0.02263573408126831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,10240,0.0255840003490448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,7168,0.01577600042025248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,7168,0.02155839999516805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,8192,0.02148800094922384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,6144,0.013704533378283182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,6144,0.020642133553822835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,7168,0.019831466674804687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,5120,0.012361600001653036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,5120,0.018942934274673463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,6144,0.017811199029286705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,4096,0.009399466713269551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,4096,0.018259199460347493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,5120,0.016124799847602844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,3584,0.008732799688975017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,3584,0.016637866695721946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,4096,0.013737600048383078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,3072,0.00788266658782959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,3072,0.015876266360282897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,3584,0.012847999731699625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,2560,0.006975999971230824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,2560,0.014971733093261719
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,3072,0.01169599990049998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,2048,0.005539200206597646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,2048,0.015001599987347921
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,2560,0.01128426690896352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,1536,0.004251733422279358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,1536,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,1536,0.009225599964459737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,1024,0.0034346667428811393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,1024,0.014447999993960061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,2048,0.010063999891281128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,768,0.003018666555484136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,768,0.014432000120480857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,768,0.007911466558774312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,512,0.0030378667016824085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,512,0.014443733294804893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,1024,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,256,0.0024661332368850707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,256,0.013056000073750814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,256,0.007474133372306823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,128,0.002162133405605952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,6144,65536,0.12514879703521728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,128,0.013479466239611307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,64,0.002164266755183538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,512,0.007479466497898102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,64,0.013434666395187377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,5120,32,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,5120,32,0.01368106702963511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,65536,0.08210453192392984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,128,0.006698666512966156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,65536,0.061161601543426515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,16384,0.026096000274022417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,16384,0.026820266246795656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,12288,0.020253866910934448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,12288,0.025033599138259886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,12288,0.02855573296546936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,10240,0.01685546636581421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,10240,0.02483946681022644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,16384,0.03585919936498006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,8192,0.014549332857131957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,8192,0.022714666525522866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,10240,0.024848000208536784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,7168,0.012988799810409546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,7168,0.02327573299407959
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,8192,0.02147093415260315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,6144,0.012037332852681477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,6144,0.02121493419011434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,7168,0.019403733809789023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,5120,0.010792533556620281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,5120,0.02172586719195048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,6144,0.017784533898035686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,4096,0.00958720048268636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,4096,0.019939200083414713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,5120,0.015752533078193666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,3584,0.01053546667098999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,3584,0.018317866325378417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,4096,0.014086400469144186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,3072,0.009517866373062133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,3072,0.016701867183049522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,3584,0.012495999534924824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,2560,0.008177066842714947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,2560,0.01664959987004598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,3072,0.012040533622105916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,2048,0.007003733515739441
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,2048,0.01649066706498464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,2560,0.011202133695284526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,1536,0.005534933507442474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,1536,0.016500266393025716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,2048,0.010365866621335347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,1024,0.004535466432571411
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,1024,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,1024,0.008396800359090168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,768,0.0038399999340375268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,768,0.014435199896494546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,1536,0.009239466985066731
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,512,0.0035264000296592714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,512,0.014499200383822122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,512,0.007439999779065449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,256,0.002593066543340683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,768,0.007960533102353413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,256,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,128,0.0022069332500298817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,128,0.014312533537546792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,128,0.00665280024210612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,64,0.0021845333278179167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,256,0.00727893312772115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,4096,32,0.0021845333278179167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,64,0.01434346636136373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,4096,32,0.013339733084042868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,65536,0.0730773369471232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,65536,0.0577184001604716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,16384,0.022366933027903237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,16384,0.025334399938583375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,5120,65536,0.124836270014445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,12288,0.01740480065345764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,12288,0.02282879948616028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,12288,0.02850666642189026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,10240,0.015197867155075073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,10240,0.022694400946299233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,16384,0.03614720106124878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,8192,0.012958932916323343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,8192,0.021386667092641195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,10240,0.0247925341129303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,7168,0.011661866307258606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,7168,0.022094933191935222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,8192,0.02129813234011332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,6144,0.01106346646944682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,6144,0.021308799584706627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,6144,0.01739733417828878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,5120,0.010070400436719258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,5120,0.01867093245188395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,7168,0.01901866594950358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,4096,0.008584533135096233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,4096,0.018372267484664917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,4096,0.013241599996884665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,3584,0.01039466659228007
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,3584,0.016528000434239708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,5120,0.016023466984430947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,3072,0.009219200412432352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,3072,0.016986666123072307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,3584,0.012506666779518127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,2560,0.00795306662718455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,2560,0.016549332936604818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,3072,0.012236799796422322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,2048,0.006911999980608622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,4096,65536,0.12564586798350016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,2048,0.01688106656074524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,1536,0.006031999985376993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,1536,0.014985600113868713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,2560,0.010849066575368245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,1024,0.005052799979845682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,1024,0.014472533265749613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,2048,0.010073600212732951
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,768,0.00393599991997083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,768,0.014696533481280008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,1024,0.007960533102353413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,1536,0.009244799613952637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,512,0.0034453332424163817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,512,0.013923199971516928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,256,0.0026389333109060925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,256,0.013920000195503235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,768,0.008040533463160197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,128,0.00225600004196167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,256,0.0071733335653940845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,512,0.0075434664885203045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,128,0.013783466815948487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,64,0.002170666555563609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3584,32,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,64,0.013128532965977987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3584,32,0.013362133502960205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,65536,0.06504746675491332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,128,0.006774400174617767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,16384,0.019751467307408652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,65536,0.052393599351247155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,16384,0.024858667453130086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,12288,0.015553067127863566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,12288,0.022808533906936646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,12288,0.02853226661682129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,10240,0.013741866747538248
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,10240,0.022986666361490885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,16384,0.03559253215789795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,8192,0.012009599804878235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,8192,0.020980266729990642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,10240,0.025624533494313557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,7168,0.011325866977373759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,7168,0.020906666914621987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,8192,0.02073600093523661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,6144,0.010233599940935771
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,6144,0.018644267320632936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,7168,0.0195743997891744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,5120,0.00941973328590393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,5120,0.018488534291585288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,6144,0.017369600137074788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,4096,0.007961600025494894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,4096,0.017017600933710735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,5120,0.016263467073440552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,3584,0.009901866316795349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,3584,0.01903146704037984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,4096,0.01362773378690084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,3072,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,3072,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,3072,0.011623467008272808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,2560,0.006772266825040181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,3584,0.013005866607030233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,2560,0.016509866714477538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,2048,0.0057429333527882894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,2048,0.016492799917856852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3584,65536,0.12489386399586995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,1536,0.005846400062243144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,1536,0.016834133863449098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,2048,0.009961600104967754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,1024,0.005021866659323374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,1024,0.014531200130780539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,2560,0.011276800433794658
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,768,0.003927466770013174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,768,0.01458133359750112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,1024,0.008409600456555684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,512,0.0033802665770053864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,512,0.014451199769973755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,768,0.008003200093905132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,256,0.0030464000999927522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,1536,0.00922879974047343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,256,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,128,0.002639999985694885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,128,0.01409386694431305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,128,0.006810666620731353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,256,0.007110400001207988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,64,0.002181333303451538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,3072,32,0.0022677332162857056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,64,0.014149333039919535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,512,0.007462400197982788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,3072,32,0.0137855996688207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,65536,0.05550080140431722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,16384,0.018642133474349974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,65536,0.046239999930063884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,16384,0.023616000016530355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,12288,0.01360319952170054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,12288,0.022308266162872313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,12288,0.02805226643880208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,10240,0.012106666962305706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,10240,0.02081706722577413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,16384,0.03544319868087768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,8192,0.010610133409500122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,8192,0.020492800076802573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,10240,0.024369066953659056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,7168,0.010193066795667012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,7168,0.019495467344919838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,8192,0.02064746618270874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,6144,0.009205333391825358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,6144,0.01862293283144633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,6144,0.017013333241144814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,5120,0.008311466872692108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,5120,0.01839573383331299
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,7168,0.01930239995320638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,4096,0.007415466507275899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,4096,0.017760000626246133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,5120,0.015380266308784484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,3584,0.006305066744486491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,3584,0.016571733355522155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,4096,0.013702399532000222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,3072,0.007218133409818013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,3072,0.01675093372662862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,3584,0.012429866194725036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,2560,0.006712533533573151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,2560,0.016590933005015053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,3072,0.011699199676513672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,2048,0.005937066674232483
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,2048,0.016886399189631144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,2560,0.010771200060844421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,1536,0.005465599894523621
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,1536,0.01600213348865509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,2048,0.010055466492970785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,1024,0.004373333354791006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,1024,0.015748266379038492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,3072,65536,0.12535040378570556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,768,0.003973333289225897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,768,0.014708266655604044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,1536,0.008828799923261006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,512,0.003458133339881897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,512,0.014451199769973755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,1024,0.008122666676839193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,256,0.002958933264017105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,256,0.014582399527231851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,512,0.007289599875609081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,128,0.0025813333690166474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,768,0.008045866588751475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,128,0.01325973371664683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,256,0.007196799914042156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,64,0.0021738665799299877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2560,32,0.0021557333568731946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,64,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2560,32,0.012999467055002847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,128,0.006724266707897187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,65536,0.04601066509882609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,65536,0.04122879902521769
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,16384,0.015622400244077048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,16384,0.022336000204086305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,12288,0.011534933249155681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,12288,0.020601600408554077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,12288,0.027698133389155073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,10240,0.01025493343671163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,10240,0.018703999121983846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,16384,0.035266133149464925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,8192,0.009723732868830364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,8192,0.021792000532150267
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,10240,0.02434879938761393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,7168,0.008796800176302593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,7168,0.02064853310585022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,8192,0.020860799153645835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,6144,0.008388266960779826
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,6144,0.0182805339495341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,7168,0.018617600202560425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,5120,0.006793599824110668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,5120,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,6144,0.017480534315109254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,4096,0.008493866523106892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,4096,0.01902079979578654
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,5120,0.015726932883262636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,3584,0.007533866663773854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,3584,0.0166101336479187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,3584,0.0127893328666687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,3072,0.005870933334032694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,3072,0.01650773286819458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,4096,0.013927466670672097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,2560,0.005238399902979533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,2560,0.016823466618855795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,3072,0.011628799637158712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,2560,0.010940800110499065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,2048,0.006320000191529592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,2048,0.016989866892496742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,1536,0.005562666555245717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,1536,0.01656000018119812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2560,65536,0.12462080319722493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,1024,0.004279466470082601
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,1024,0.014674133062362671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,1536,0.009250133236249288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,1024,0.008372267087300617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,768,0.0037962667644023894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,768,0.014615466197331747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,512,0.003502933432658514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,2048,0.009964799880981446
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,512,0.014506666858990987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,256,0.002639999985694885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,256,0.013885866602261862
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,256,0.007175466914971669
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,512,0.0070933332045873005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,768,0.007952000200748443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,128,0.002551466723283132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,128,0.014519466956456503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,64,0.0022463999688625337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,128,0.006684799989064534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,2048,32,0.002183466653029124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,64,0.014537599682807923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,2048,32,0.013708800077438354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,65536,0.035997867584228516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,65536,0.034756267070770265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,16384,0.012115200360616047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,16384,0.024035199483235677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,12288,0.01516800026098887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,12288,0.02164693276087443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,12288,0.027777065833409626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,10240,0.012659200032552085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,10240,0.020849066972732543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,16384,0.035087998708089194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,8192,0.01213759978612264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,8192,0.02109760046005249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,8192,0.02066453297932943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,7168,0.010392533739407857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,7168,0.018948266903559365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,10240,0.024615466594696045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,6144,0.009326933821042379
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,6144,0.0181877334912618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,7168,0.019027199347813925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,5120,0.008356266220410665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,5120,0.016595199704170227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,6144,0.017566933234532674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,4096,0.00727893312772115
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,4096,0.018312533696492515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,5120,0.015307733416557312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,3584,0.006387199958165486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,3584,0.016484266519546507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,4096,0.013778133193651834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,3072,0.005672533313433329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,3072,0.0168778657913208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,3584,0.01242026686668396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,2560,0.0050794666012128195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,2560,0.016551466782887776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,3072,0.012156800429026286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,2048,0.006030933558940887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,2048,0.015597866972287497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,2560,0.010784000158309937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,1536,0.004554666578769684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,1536,0.015980799992879234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,2048,0.0100874662399292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,1024,0.0043935999274253845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,1024,0.01462399959564209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,1536,0.008948266506195068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,768,0.00383146678407987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,768,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,1024,0.00848533312479655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,512,0.003489066660404205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,768,0.0074869334697723385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,512,0.014571733276049294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,256,0.0027434666951497394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,256,0.013392000397046407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,2048,65536,0.12350506782531738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,128,0.002327466756105423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,512,0.0071487997968991595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,128,0.014425599575042724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,64,0.002266666789849599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,256,0.007527466615041096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,128,0.006354133288065593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1536,32,0.00225600004196167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,64,0.01349440018335978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1536,32,0.013892267147699991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,65536,0.025179733832677204
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,16384,0.013818666338920593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,65536,0.02831679979960124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,16384,0.02063039938608805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,12288,0.011361066500345867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,12288,0.019008000691731773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,12288,0.028058665990829467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,10240,0.009998933474222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,10240,0.018611200650533042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,16384,0.03499093453089396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,8192,0.008243200182914735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,8192,0.018708266814549766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,10240,0.024382932980855306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,7168,0.00759680022795995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,7168,0.019062399864196777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,8192,0.02071466644605001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,6144,0.006724266707897187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,6144,0.016571733355522155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,7168,0.019106133778889974
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,5120,0.0063360000650088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,5120,0.016595199704170227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,6144,0.017492266496022542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,4096,0.005890133480230967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,4096,0.01842986742655436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,5120,0.015719466408093772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,3584,0.005497600138187409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,3584,0.01678826610247294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,4096,0.013246933619181315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,3072,0.005510400235652924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,3072,0.016515200336774193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,3584,0.012836266557375589
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,2560,0.005227733155091604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,2560,0.01566933294137319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,3072,0.011687466502189636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,2048,0.006265600025653839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,2048,0.015878400206565856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,2560,0.010870400071144103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,1536,0.005054933329423269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,1536,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,2048,0.010066133737564088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,1024,0.003894400099913279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,1024,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1536,65536,0.12190933227539062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,768,0.0035114665826161706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,768,0.014605866869290671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,1536,0.009161600470542907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,512,0.0033813332517941795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,1024,0.00832533339659373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,512,0.014442666371663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,768,0.007916800181070964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,256,0.0025941332181294756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,256,0.014502400159835815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,128,0.002567466596762339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,128,0.014416000247001648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,256,0.006687999765078227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,512,0.007594666878382365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,64,0.002164266755183538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,1024,32,0.0021578667064507803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,64,0.013983999689420065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,128,0.00673280010620753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,1024,32,0.014153599739074707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,65536,0.021025067567825316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,16384,0.011394133170445759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,65536,0.0333024005095164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,16384,0.020363734165827433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,12288,0.009634133179982502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,12288,0.018602667252222697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,12288,0.02804159919420878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,10240,0.008402132987976074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,10240,0.018806399901707967
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,16384,0.034969600041707356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,8192,0.007109333574771881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,8192,0.018769067525863648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,10240,0.024385066827138265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,7168,0.006401066482067108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,7168,0.018614399433135986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,8192,0.020424532890319824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,6144,0.006678399940331777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,6144,0.018548266092936198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,7168,0.019036799669265747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,5120,0.005529599885145823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,5120,0.016945066054662068
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,6144,0.01699413259824117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,4096,0.0060127998391787205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,4096,0.017416532834370932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,5120,0.015709867080052696
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,3584,0.00584853341182073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,3584,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,4096,0.01325440009435018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,3072,0.0054666668176651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,3072,0.016495999693870545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,1024,65536,0.12028480370839437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,2560,0.0055871998270352686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,2560,0.0164682666460673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,3584,0.012525866429011026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,2048,0.0043029333154360454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,2048,0.014634666840235391
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,3072,0.012010666728019714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,1536,0.0046741331617037455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,2560,0.010427733262379963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,1536,0.015177599589029946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,1024,0.003874133278926214
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,1024,0.015040000279744467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,2048,0.009983999530474345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,768,0.0034346667428811393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,1536,0.008775466680526733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,1024,0.008007466793060303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,512,0.0031008000175158186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,768,0.014513066411018372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,512,0.014547200004259745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,256,0.0026805333793163298
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,256,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,768,0.00790826678276062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,512,0.007584000130494435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,128,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,128,0.013927466670672097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,256,0.007271466652552287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,64,0.002238933245340983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,128,0.00628053347269694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,768,32,0.002219733347495397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,64,0.014538666605949402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,768,32,0.013412266969680786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,65536,0.02844799955685933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,16384,0.008351999521255492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,65536,0.027030400435129803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,16384,0.01876266598701477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,12288,0.0075647999842961625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,12288,0.018637865781784058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,12288,0.027908267577489217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,10240,0.006724266707897187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,16384,0.03466453154881795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,10240,0.01871253252029419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,8192,0.00573333352804184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,8192,0.019661867618560792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,10240,0.024370133876800537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,7168,0.00547733356555303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,7168,0.018869332472483315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,8192,0.021118932962417604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,6144,0.005946666498978933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,6144,0.017427200078964235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,7168,0.01864746610323588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,5120,0.0055071999629338585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,5120,0.018323200941085815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,6144,0.01738026738166809
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,4096,0.005881600081920624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,4096,0.016561067104339598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,5120,0.015066666404406228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,3584,0.00552106648683548
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,3584,0.0164192001024882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,4096,0.01383039951324463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,3072,0.0053610667586326596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,3072,0.016009599963823954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,3584,0.012451199690500896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,2560,0.0051146666208903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,2560,0.016286933422088624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,3072,0.01167146662871043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,768,65536,0.12030613422393799
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,2048,0.004643199841181437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,2560,0.010788266857465107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,2048,0.015495466192563376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,1536,0.0038304001092910765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,1536,0.014603733023007711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,1024,0.0034815999368826545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,1024,0.014499200383822122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,1536,0.008782933155695598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,2048,0.009998933474222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,768,0.003049599876006444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,1024,0.00792746643225352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,768,0.014487466216087342
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,512,0.0025973332424958544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,512,0.014453333616256715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,256,0.0025727999707063037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,768,0.0075541332364082335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,256,0.013801599542299906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,128,0.002217599997917811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,512,0.007496533294518788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,128,0.014203733205795288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,256,0.006682666639486949
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,64,0.0021589333812395734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,512,32,0.0021802666286627453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,128,0.006275199850400289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,64,0.014151466886202493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,65536,0.016503467162450155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,512,32,0.013331199685732523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,16384,0.005860266586144766
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,65536,0.022371200720469157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,16384,0.018687999248504637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,12288,0.005103999873002371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,12288,0.016924800475438435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,12288,0.027999999125798543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,10240,0.00551146666208903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,10240,0.017195733388264973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,16384,0.03466773430506388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,8192,0.005596800148487091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,8192,0.01744426687558492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,10240,0.024778666098912557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,7168,0.005481599768002828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,7168,0.016523733735084534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,8192,0.02023786703745524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,6144,0.0050784001747767125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,6144,0.016701867183049522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,7168,0.01939520041147868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,5120,0.005046399931112925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,6144,0.01698346734046936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,5120,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,4096,0.0046410664916038515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,4096,0.016566399733225504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,512,65536,0.11993920008341472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,5120,0.01532906691233317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,3584,0.0051018665234247845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,4096,0.01386666695276896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,3584,0.0165503998597463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,3072,0.004666666686534882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,3072,0.015266133348147073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,2560,0.005017599960168203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,2560,0.015948800245920818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,3584,0.012505599856376648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,3072,0.012027733524640401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,2048,0.0046410664916038515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,2048,0.014589866995811463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,1536,0.0038272000849246977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,2560,0.010634666681289673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,1536,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,1024,0.003010133405526479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,1024,0.014526933431625366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,2048,0.009971200426419576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,1536,0.008776533603668212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,768,0.003019733230272929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,768,0.014517333110173544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,512,0.002631466587384542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,1024,0.007979733248551685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,768,0.008217599987983704
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,512,0.014293332894643148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,256,0.0022698665658632917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,512,0.007119999825954437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,256,0.01340053379535675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,128,0.002295466760794322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,256,0.007091199855009715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,128,0.014034133156140646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,128,0.006346666812896728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,64,0.0022762666145960493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,64,0.01292693316936493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,256,32,0.002446933339039485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,65536,0.010107733805974324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,256,32,0.013962666193644205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,65536,0.019612799088160195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,16384,0.005998933315277099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,16384,0.018524799744288126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,16384,0.035130667686462405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,12288,0.005157333115736643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,12288,0.01688106656074524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,10240,0.005545599758625031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,10240,0.016938666502634682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,256,65536,0.1196991999944051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,8192,0.005172266562779745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,8192,0.017258665959040322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,10240,0.024639999866485594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,12288,0.027723733584086103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,7168,0.00506986677646637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,7168,0.01728106737136841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,8192,0.020572799444198608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,6144,0.004999466737111409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,6144,0.016515200336774193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,5120,0.00425600012143453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,5120,0.01681173245112101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,7168,0.01917120019594828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,4096,0.004634666442871094
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,6144,0.01733013391494751
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,4096,0.01653866668542226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,3584,0.0051018665234247845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,5120,0.01532799998919169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,3584,0.015808000167210897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,3072,0.004731733103593191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,3072,0.015012266238530478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,4096,0.013359999656677246
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,2560,0.004621866842110952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,3584,0.012523733576138816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,2560,0.0167413334051768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,2048,0.004299733539422353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,3072,0.011664000153541566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,2048,0.014989866813023885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,1536,0.0038431999584039056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,2560,0.010846933722496033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,1536,0.014522666732470194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,1024,0.0030773334205150605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,2048,0.010397866368293762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,1024,0.01467626690864563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,1536,0.008785067001978557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,65536,0.11950186888376872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,768,0.003035733352104823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,512,0.002690133452415466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,768,0.013246933619181315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,1024,0.00832426647345225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,256,0.002204799900452296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,512,0.014447999993960061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,768,0.007612800101439159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,512,0.007538133362929027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,256,0.013512532909711203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,128,0.0021589333812395734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,256,0.006771199901898702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,64,0.0018207999567190808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,128,0.012755200266838074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,128,32,0.0017984000345071156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,4,128,128,0.007108266651630402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,64,0.014123732844988504
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,128,32,0.01290773351987203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,65536,0.0067445332805315655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,16384,0.0054506664474805195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,65536,0.020838399728139244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,12288,0.005049600203831991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,10240,0.0050346667567888895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,16384,0.017035732666651406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,12288,0.016662399967511496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,8192,0.005038933455944061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,7168,0.005046399931112925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,10240,0.01662506659825643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,8192,0.017398399114608765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,6144,0.004637866715590159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,7168,0.01753066579500834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,5120,0.0042303999265035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,4096,0.0046410664916038515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,6144,0.01649386684099833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,5120,0.016669867436091106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,3584,0.004604800045490265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,4096,0.01656426688035329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,3072,0.004604800045490265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,2560,0.004712533454100291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,3072,0.015233066678047181
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,3584,0.014485333363215128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,2048,0.004212266703446706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,2560,0.016083199779192606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,1536,0.003454933315515518
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,2048,0.014494933684666953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,1024,0.0029866665601730345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,1536,0.01498133341471354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,768,0.002661333233118057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,512,0.0024106666445732118
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,1024,0.014450132846832275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,768,0.014548266927401224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,256,0.0022261333962281544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,128,0.0022304000953833262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,512,0.01290880044301351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,256,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,64,0.0018239999810854596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,128,0.014019200205802917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,64,32,0.0018325333793958027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,64,0.014122666915257773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,65536,0.0067775999506314594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,64,32,0.013825066884358726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,16384,0.005029333134492239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,65536,0.018917334079742432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,12288,0.004789333542188009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,10240,0.004704000055789947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,16384,0.017489065726598106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,12288,0.017869865894317626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,8192,0.004653866589069367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,10240,0.016545066237449647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,7168,0.005054933329423269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,6144,0.0052149335543314615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,8192,0.01665493349234263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,7168,0.017066667477289833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,5120,0.004227200150489807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,4096,0.004231466849644979
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,6144,0.016587733229001363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,5120,0.01657600005467733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,3584,0.004670933385690053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,3072,0.004791466891765595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,4096,0.017036799589792886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,3584,0.01452906628449758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,2560,0.004278400043646494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,2048,0.003864533454179764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,3072,0.015029333035151162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,2560,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,1536,0.003818666686614355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,1024,0.0030464000999927522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,2048,0.014727466305096946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,1536,0.014575999975204468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,768,0.0026730666557947796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,512,0.0025727999707063037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,1024,0.014507733782132468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,768,0.013035733501116434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,256,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,128,0.0022133332987626392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,512,0.014069333672523499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,256,0.012587733070055642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,64,0.002054399996995926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,4,32,32,0.0019999999552965165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,128,0.014227199554443359
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,64,0.012578133742014566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,4,32,32,0.013805866241455078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,16384,0.17601067225138348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,16384,0.3275701204935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,12288,0.23486186663309733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,12288,0.1386016050974528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,10240,0.19965333938598634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,10240,0.11794666449228923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,8192,0.16095679601033527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,8192,0.10044266382853191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,8192,0.07171520392100016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,7168,0.14146560033162434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,7168,0.08303146362304688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,10240,0.07767679691314697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,6144,0.12470933596293132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,6144,0.07255679766337077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,12288,0.09653333028157553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,5120,0.10380586783091228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,5120,0.06370346546173096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,6144,0.049838932355244954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,4096,0.08808746337890624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,4096,0.05346986850102743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,7168,0.05996479988098145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,3584,0.06989866892496745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,3584,0.04888213475545247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,5120,0.048232531547546385
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,3072,0.06032853523890177
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,3072,0.04469546476999919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,16384,0.13203840255737304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,2560,0.050514133771260586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,2560,0.04031146764755249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,3584,0.036115201314290364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,2048,0.04084906578063965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,2048,0.03460586468378703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,4096,0.03861120144526164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,1536,0.03190400004386902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,1536,0.029310933748881024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,2560,0.026723200082778932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,1024,0.022190932432810465
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,1024,0.02482666571935018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,3072,0.03132479985555013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,768,0.018066134055455527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,768,0.022681599855422972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,1536,0.01946559945742289
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,512,0.012903466820716858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,512,0.019244800011316933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,2048,0.023858133951822916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,256,0.007761066655317943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,256,0.01672640045483907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,1024,0.01534293293952942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,128,0.004711466530958811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,128,0.016472533345222473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,768,0.0137472003698349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,64,0.004257066547870636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,64,0.01608746647834778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,65536,32,0.004218666752179464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,65536,32,0.01579200029373169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,65536,0.3193056106567383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,65536,0.17605867385864257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,128,0.008478933572769165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,16384,0.09057919979095459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,16384,0.06161493460337321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,256,0.009564800063769023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,12288,0.06574186484018961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,12288,0.04501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,65536,512,0.011322666207949321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,10240,0.056051198641459146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,10240,0.03997759819030762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,10240,0.03179626663525899
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,8192,0.046514133612314865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,8192,0.03474986553192139
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,12288,0.0388703982035319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,7168,0.03607573509216309
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,7168,0.0321557343006134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,16384,0.050348798433939614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,6144,0.03133866588274638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,6144,0.028981333971023558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,8192,0.027926399310429888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,5120,0.02677546739578247
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,5120,0.026818132400512694
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,7168,0.025084799528121947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,4096,0.021964800357818604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,4096,0.024359466632207234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,6144,0.022361600399017335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,3584,0.019620267550150554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,3584,0.023114667336146037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,4096,0.017787732680638633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,3072,0.01697280009587606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,3072,0.021415466070175172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,5120,0.019000534216562906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,2560,0.01461120049158732
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,2560,0.020615466435750327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,2560,0.013668266932169595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,2048,0.012403200070063274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,2048,0.01876693367958069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,3072,0.014899200201034546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,1536,0.009947733084360758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,1536,0.01856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,3584,0.015733333428700765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,1024,0.007482666770617168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,1024,0.015078399578730264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,1024,0.008794666330019633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,768,0.0063285330931345625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,768,0.014630400141080222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,1536,0.010411733388900756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,512,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,512,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,2048,0.012075733145078022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,256,0.0030016000072161358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,256,0.013980799913406372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,768,0.008277333279450735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,128,0.0026410666604836782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,128,0.014437333742777506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,512,0.007919999957084655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,64,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,64,0.014034133156140646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,16384,32,0.0021920000513394673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,16384,32,0.014071466525395713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,128,0.006781866649786632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,65536,0.23204053243001305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,65536,0.13787733713785807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,16384,0.06842346986134848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,16384,0.04938986698786418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,256,0.007530666887760162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,12288,0.04985280036926269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,12288,0.04068906704584758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,12288,0.030002133051554365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,10240,0.04254613320032756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,10240,0.03734933137893677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,16384,0.037514666716257736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,8192,0.03552213509877523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,8192,0.034092799822489424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,10240,0.025886933008829754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,7168,0.03134933312733968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,7168,0.030872533718744915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,8192,0.02209279934565226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,6144,0.027346134185791016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,6144,0.028849067290623982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,7168,0.02018666664759318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,5120,0.024566400051116943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,5120,0.026754132906595868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,6144,0.01844480037689209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,4096,0.020547199249267577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,4096,0.02653333346048991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,5120,0.016717867056528727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,3584,0.018552533785502114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,3584,0.0206496000289917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,4096,0.014492799838383993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,3072,0.015921066204706825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,3072,0.020660267273585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,3584,0.013685333728790283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,2560,0.014206932981808982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,2560,0.018716800212860107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,2560,0.011238400141398113
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,2048,0.012650666634241739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,2048,0.01858666737874349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,3072,0.01276693344116211
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,1536,0.0083146666487058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,1536,0.01643946667512258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,1536,0.009171199798583985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,1024,0.006301866471767425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,1024,0.014685866236686707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,2048,0.010375466942787171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,768,0.004193066557248434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,768,0.014518400033315023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,1024,0.008205866813659668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,512,0.003479466587305069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,512,0.014579199751218162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,768,0.007914666831493378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,256,0.003019733230272929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,256,0.01411733329296112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,256,0.0071285332242647815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,128,0.002583466718594233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,512,0.007490133245786031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,128,0.013688533504803976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,64,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,64,0.012935466567675271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,12288,32,0.002209066599607468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,12288,32,0.013665067156155905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,65536,0.19321920077006022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,128,0.006733866532643636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,65536,0.11944320201873779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,16384,0.05885013341903687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,16384,0.04276586771011352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,16384,0.03714453379313151
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,12288,0.04136213461558024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,12288,0.03701866865158081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,12288,0.02930026650428772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,10240,0.03583039840062459
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,10240,0.03399466673533122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,16384,65536,0.1662442684173584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,8192,0.029044266541798907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,8192,0.030784000953038532
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,10240,0.025715200106302898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,7168,0.027266132831573486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,7168,0.028910932938257854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,8192,0.021705599625905354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,6144,0.023571199178695677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,6144,0.02672426700592041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,7168,0.01996799906094869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,5120,0.020713599522908528
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,5120,0.025571199258168538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,12288,65536,0.13301866849263508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,4096,0.01690773367881775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,4096,0.020932267109553017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,6144,0.018501333395640054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,3584,0.015836800138155617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,3584,0.022487467527389525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,5120,0.0166101336479187
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,3072,0.014883200327555338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,4096,0.014443733294804893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,3072,0.020593067010243736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,2560,0.013335466384887695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,2560,0.01939520041147868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,3584,0.013012267152468362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,2048,0.010136533776919048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,2048,0.018667733669281004
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,2560,0.01130986710389455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,1536,0.008262399832407634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,1536,0.01676586667696635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,3072,0.012114133437474568
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,1024,0.006774400174617767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,1024,0.016472533345222473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,1536,0.009289600451787313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,768,0.005145599941412607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,768,0.014572800199190775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,1024,0.00844693382581075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,512,0.004301866888999939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,512,0.014498133460680643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,2048,0.010006399949391682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,256,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,256,0.01304213305314382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,512,0.007613866527875264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,128,0.00223786657055219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,128,0.013474133610725404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,768,0.007977599898974102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,64,0.002243199944496155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,64,0.013779200116793313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,10240,32,0.00223786657055219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,10240,32,0.013015466928482055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,65536,0.15753173828125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,128,0.0067658667763074234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,256,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,65536,0.09685546557108561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,16384,0.04719039996465047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,16384,0.03587626616160075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,12288,0.034584534168243405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,12288,0.031231999397277832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,12288,0.028827732801437377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,10240,0.029706666866938274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,10240,0.028586665789286297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,16384,0.036183468500773114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,8192,0.022887466351191203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,8192,0.024887466430664064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,10240,0.02553173303604126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,7168,0.01978666583697001
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,7168,0.023605332771937052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,8192,0.02187839945157369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,6144,0.017892267306645712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,6144,0.022478934129079184
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,10240,65536,0.12768853505452474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,5120,0.015341867009798685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,5120,0.02155839999516805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,7168,0.01982933282852173
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,4096,0.012462932864824932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,4096,0.018604799111684164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,6144,0.018309332927068076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,3584,0.011542399724324543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,3584,0.019064533710479736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,5120,0.016313599546750386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,3072,0.010364799698193868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,3072,0.018686934312184652
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,4096,0.014510933558146158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,2560,0.009147733449935913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,2560,0.016509866714477538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,3584,0.012888532876968384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,2048,0.007788800199826558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,2048,0.01565439999103546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,2560,0.01079360047976176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,1536,0.006301866471767425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,1536,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,3072,0.01202880044778188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,1024,0.004275199770927429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,1024,0.014556800325711569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,2048,0.010020266969998677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,768,0.0034624000390370687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,768,0.01454080045223236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,1024,0.00816319982210795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,512,0.0034346667428811393
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,512,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,1536,0.009147733449935913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,256,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,256,0.013623467087745667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,768,0.007971199850241344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,128,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,128,0.012918399771054587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,256,0.007062399884064992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,64,0.0021568000316619872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,64,0.013438933094342551
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,8192,32,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,8192,32,0.013168000181516013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,128,0.007119999825954437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,65536,0.1371338685353597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,65536,0.08706560134887695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,16384,0.039918935298919676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,512,0.007531733314196269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,16384,0.03539520104726156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,12288,0.03090133269627889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,12288,0.028961066404978437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,12288,0.028869332869847615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,10240,0.026630399624506633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,10240,0.027670399347941084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,16384,0.03586346705754598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,8192,0.02191466689109802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,8192,0.024528000752131143
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,10240,0.025623466571172076
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,7168,0.01956053376197815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,7168,0.022553600867589316
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,8192,0.02159573237101237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,6144,0.01597546637058258
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,6144,0.02102400064468384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,7168,0.02030293345451355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,5120,0.014308266838391624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,5120,0.020669867595036827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,6144,0.01823893388112386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,4096,0.011377066373825073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,4096,0.020216532548268638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,5120,0.016594133774439492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,3584,0.010601600011189777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,3584,0.018582399686177573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,4096,0.014097066720326743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,3072,0.009250133236249288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,3072,0.01864853302637736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,3584,0.012990933656692506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,2560,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,2560,0.01656000018119812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,3072,0.011666133006413778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,2048,0.007180800040562947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,2048,0.016339199741681416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,2560,0.011268267035484314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,1536,0.005779199798901876
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,1536,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,2048,0.010018133123715718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,1024,0.004144000013669332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,1024,0.014597333470980325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,1536,0.00921066701412201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,768,0.003483733286460241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,768,0.01444906691710154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,1024,0.007974400122960409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,512,0.003044266750415166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,512,0.014625066518783569
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,768,0.0075103998184204105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,256,0.0025973332424958544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,256,0.014473600188891092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,256,0.007118933399518331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,128,0.002179199953873952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,128,0.013705600301424661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,512,0.007211733361085255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,64,0.0021429332594076794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,64,0.014424533645311991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,7168,32,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,7168,32,0.014359466234842935
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,128,0.0070826664566993715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,65536,0.11840213139851888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,65536,0.07974293231964111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,16384,0.03694506486256917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,16384,0.03269333243370056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,8192,65536,0.12571840286254882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,12288,0.027334400018056232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,12288,0.028943999608357744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,16384,0.036235733826955156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,10240,0.0240831991036733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,10240,0.027433600028355914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,12288,0.028485333919525145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,8192,0.019197867314020792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,8192,0.022833067178726196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,8192,0.02185813387235006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,7168,0.017452800273895265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,7168,0.020747733116149903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,10240,0.025727999210357667
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,6144,0.016151466965675355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,6144,0.02071040074030558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,6144,0.018199467658996583
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,5120,0.014293332894643148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,5120,0.020257065693537392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,7168,0.01983786622683207
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,4096,0.010694400469462077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,4096,0.01872640053431193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,4096,0.013358933726946512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,3584,0.009477333227793375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,3584,0.018652800718943277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,5120,0.01583573321501414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,3072,0.00877333382765452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,3072,0.01676373283068339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,3584,0.012852266430854797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,2560,0.007590400179227193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,2560,0.015969066818555196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,3072,0.012185600399971009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,2048,0.00690773328145345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,2048,0.014932266871134438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,2560,0.010802132884661357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,1536,0.004770133395989736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,1536,0.014484266440073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,2048,0.0101173331340154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,1024,0.003982933362325033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,1024,0.014643200238545737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,1024,0.008441600203514098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,768,0.003606399893760681
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,768,0.014662399888038635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,7168,65536,0.12559999624888102
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,512,0.0031360000371932983
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,512,0.01461013356844584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,1536,0.00913706620534261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,256,0.002573866645495097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,256,0.01411733329296112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,256,0.0070816000302632645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,128,0.002216533323129018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,128,0.013371733824412027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,128,0.007076266904671986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,512,0.0076896001895268755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,64,0.002149333308140437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,6144,32,0.002179199953873952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,32,0.013343999783198038
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,6144,64,0.013412266969680786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,65536,0.10149866739908855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,768,0.007969066500663757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,65536,0.07096213499704997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,16384,0.03207359910011291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,16384,0.03161493341128031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,12288,0.02355946699778239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,12288,0.02691199978192647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,12288,0.028514132897059126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,10240,0.020220800240834554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,10240,0.024718934297561647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,16384,0.035981865723927815
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,8192,0.018578133980433145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,8192,0.021670399109522502
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,10240,0.025148799022038777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,7168,0.01569706698258718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,7168,0.020623999834060668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,8192,0.021499733130137123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,6144,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,6144,0.01954666574796041
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,7168,0.019426133235295615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,5120,0.012011733651161195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,5120,0.019127466281255088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,6144,0.01777600049972534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,4096,0.009495466947555542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,4096,0.01771519978841146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,5120,0.015738667050997416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,3584,0.008776533603668212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,3584,0.016613333423932394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,4096,0.01334933340549469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,3072,0.007895466685295106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,3072,0.016242133577664693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,3584,0.012874666849772134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,2560,0.007150933146476746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,2560,0.014669866363207499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,3072,0.012055466572443645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,2048,0.0055061335364977515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,2048,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,2560,0.011204266548156738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,1536,0.004632533093293508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,1536,0.014482133587201438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,6144,65536,0.12485973040262859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,1024,0.0036757332583268487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,1024,0.01458560029665629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,1536,0.009179733196894328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,768,0.0034442665676275887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,768,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,1024,0.007981866598129272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,512,0.0026933332284291584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,512,0.014719999829928079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,2048,0.009949866930643718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,256,0.0025994665920734406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,256,0.01407360037167867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,768,0.00790719985961914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,128,0.002183466653029124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,128,0.013152000308036805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,256,0.007092266778151194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,64,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,64,0.013223466277122498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,128,0.006714666883150737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,512,0.007463466624418895
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,5120,32,0.0021802666286627453
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,5120,32,0.013412266969680786
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,65536,0.08093012968699137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,65536,0.06025600035985311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,16384,0.02632960081100464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,16384,0.026925865809122724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,12288,0.020167466004689535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,12288,0.024893865982691447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,12288,0.02884053389231364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,10240,0.017473065853118898
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,10240,0.022674133380254112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,16384,0.03580373525619507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,8192,0.014267733693122864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,8192,0.022741333643595377
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,10240,0.02518933415412903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,7168,0.013609600067138673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,7168,0.023042132457097374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,8192,0.02118399937947591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,6144,0.011960533261299134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,6144,0.022388267517089843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,7168,0.01943999926249186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,5120,0.01107199986775716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,5120,0.02067626714706421
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,6144,0.01734293301900228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,4096,0.00955733358860016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,4096,0.02064746618270874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,5120,0.015745066603024802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,3584,0.010714667042096455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,3584,0.018518400192260743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,4096,0.013676800330479941
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,3072,0.009989333152770997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,3072,0.018590933084487914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,3584,0.012923733393351237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,2560,0.008374399940172831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,2560,0.016598400473594666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,3072,0.011715199549992878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,2048,0.00829013337691625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,2048,0.016665599743525186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,2560,0.011236266295115153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,1536,0.0059338668982187905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,1536,0.015270400047302245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,2048,0.009998933474222819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,1536,0.009514666597048442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,1024,0.005050666630268097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,1024,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,768,0.004215466479460398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,768,0.014498133460680643
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,768,0.00798826664686203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,512,0.0034261333445707956
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,512,0.01428053379058838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,1024,0.00831573357184728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,256,0.002632533262173335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,256,0.014422399799029031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,256,0.007131733496983846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,128,0.0022314667701721193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,512,0.007518933216730754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,128,0.014403200149536133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,64,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,128,0.007099733253320058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,64,0.013739732901255288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,4096,32,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,4096,32,0.013724799950917563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,65536,0.07285439968109131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,65536,0.05480853319168091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,16384,0.02283733288447062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,16384,0.026692267258961993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,5120,65536,0.12490986982981364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,12288,0.017052799463272095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,12288,0.022709333896636964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,12288,0.028501333793004353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,10240,0.015053866306940713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,10240,0.022399999698003135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,16384,0.03588693141937256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,8192,0.01297599971294403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,8192,0.02236586610476176
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,10240,0.024855466683705647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,7168,0.012317867080370585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,7168,0.0207370658715566
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,8192,0.021223467588424683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,6144,0.011367467045783997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,6144,0.020925867557525634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,7168,0.01943146586418152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,5120,0.010030933221181234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,5120,0.019619200627009073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,6144,0.016988799969355265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,4096,0.009201066692670186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,4096,0.018931200106938682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,5120,0.015731199582417806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,3584,0.0105621337890625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,3584,0.0184714674949646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,4096,0.013384532928466798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,3072,0.010055466492970785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,3072,0.01872640053431193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,4096,65536,0.12534613609313966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,2560,0.008002133170763651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,2560,0.016619732975959776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,3584,0.012497066458066305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,2048,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,2048,0.01653866668542226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,3072,0.01186346709728241
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,1536,0.006783999999364217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,1536,0.01584106683731079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,2560,0.010925867160161336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,1024,0.005141333242257436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,1024,0.01460906664530436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,2048,0.010026666522026061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,768,0.004242133100827535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,768,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,1536,0.009258666634559631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,512,0.003565866748491923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,512,0.014503467082977294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,512,0.007520000139872233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,256,0.0026591998835404714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,768,0.008224000036716462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,256,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,128,0.0022570667167504626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,1024,0.008370133241017659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,128,0.014442666371663412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,64,0.00225600004196167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,64,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3584,32,0.0021664001047611236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,256,0.007119999825954437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,128,0.006737066805362702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3584,32,0.01439786652723948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,65536,0.06381119887034098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,65536,0.05129493474960327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,16384,0.020362667242685952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,16384,0.02577173312505086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,12288,0.015375999609629312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,12288,0.022825600703557332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,12288,0.028163200616836546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,10240,0.013697066903114319
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,10240,0.021721599499384563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,16384,0.03545173406600952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,8192,0.012248532970746358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,8192,0.020652800798416138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,10240,0.024887466430664064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,7168,0.010915199915568035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,7168,0.020770132541656494
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,8192,0.020757333437601725
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,6144,0.010248532891273499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,6144,0.019197867314020792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,7168,0.01917546590169271
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,5120,0.009125333031018574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,5120,0.018802134195963542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,6144,0.017033600807189943
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,4096,0.007971199850241344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,4096,0.018681599696477254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,5120,0.01589120030403137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,3584,0.009432533383369445
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,3584,0.018776534001032512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,4096,0.01325759987036387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,3072,0.007913599908351897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,3072,0.018662399053573607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,3072,0.01200213332970937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,2560,0.007336533566315968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,2560,0.016743467251459757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,3584,0.013081600268681845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3584,65536,0.1258133331934611
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,2048,0.006523733337720235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,2048,0.016682666540145875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,1536,0.006857599814732869
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,1536,0.01579200029373169
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,2560,0.010875733693440755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,1024,0.005019733309745788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,1024,0.014579199751218162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,1024,0.00790719985961914
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,768,0.004216533402601878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,1536,0.009244799613952637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,768,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,512,0.0034826666116714476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,512,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,2048,0.010082133611043294
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,256,0.0031487998863061273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,256,0.01455466647942861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,512,0.007551999886830647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,128,0.0025792000194390613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,128,0.013716266552607218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,256,0.007229866584142049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,64,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,768,0.008021333316961924
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,3072,32,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,64,0.013780267039934794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,3072,32,0.013553067048390707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,65536,0.05562880039215088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,128,0.00668693333864212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,16384,0.018150399128595986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,65536,0.04498026768366496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,16384,0.02378666599591573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,12288,0.014144000411033631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,12288,0.021222400665283202
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,12288,0.028141866127649944
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,10240,0.012803199887275695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,10240,0.020692267020543418
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,16384,0.03583360115687052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,8192,0.010878933469454448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,8192,0.018915200233459474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,10240,0.024514132738113405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,7168,0.009988266229629516
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,7168,0.018702934185663857
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,8192,0.021269333362579346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,6144,0.009335466225941976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,6144,0.019398399194081626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,7168,0.018985599279403687
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,5120,0.00837546686331431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,5120,0.018605866034825645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,6144,0.01758079926172892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,4096,0.00730453332265218
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,4096,0.01672640045483907
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,5120,0.015382400155067444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,3584,0.006278400123119354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,3584,0.018600533405939736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,4096,0.013767466942469279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,3072,0.00772266685962677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,3072,0.016705065965652466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,3584,0.012403200070063274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,2560,0.00709440012772878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,2560,0.016549332936604818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,2560,0.010802132884661357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,2048,0.007521066566308339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,2048,0.01649493376413981
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,3072,0.012149332960446676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,1536,0.006433066725730896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,1536,0.01514346698919932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,2048,0.009966933727264404
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,1024,0.005074133475621542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,1536,0.009221333265304565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,1024,0.014533332983652749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,768,0.00425493319829305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,768,0.014521599809328715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,3072,65536,0.12527146339416503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,512,0.003532800078392029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,512,0.014616533120473226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,768,0.007502933343251546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,256,0.0026549334327379864
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,256,0.014245333274205527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,512,0.0075882668296496075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,128,0.002628266563018163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,1024,0.00839466651280721
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,128,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,64,0.002217599997917811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,64,0.014491732915242514
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,128,0.006361599763234456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2560,32,0.0021375998854637145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,256,0.006702933212121327
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2560,32,0.014419200023015341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,65536,0.045324798425038657
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,16384,0.015373866756757101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,65536,0.04023146629333496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,16384,0.022760534286499025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,12288,0.01179093321164449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,12288,0.020802134275436403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,12288,0.027988266944885255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,10240,0.010520533720652262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,10240,0.020483199755350748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,16384,0.03526826699574788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,8192,0.0091839998960495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,8192,0.02134400010108948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,10240,0.024474666515986124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,7168,0.008919466535250347
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,7168,0.020793600877126058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,8192,0.02066133419672648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,6144,0.007768533130486806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,6144,0.01767359972000122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,7168,0.019179733594258626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,5120,0.007002666592597961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,5120,0.01885226567586263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,6144,0.01730239987373352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,4096,0.007996800045172375
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,4096,0.018623999754587808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,5120,0.015453867117563882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,3584,0.0076885332663853955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,3584,0.019048533837000527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,4096,0.013255467017491659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,3072,0.006316799918810527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,3072,0.01806079943974813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,3584,0.012526933352152506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,2560,0.006001066664854685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,2560,0.017284266153971353
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,3072,0.011598933736483257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,2048,0.007580799857775371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,2048,0.016482133666674295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,2560,0.011028266946474711
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,1536,0.006449066599210103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,1536,0.016502400239308677
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2560,65536,0.12497920195261639
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,1024,0.004846933484077454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,1024,0.014617600043614707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,2048,0.00992746651172638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,768,0.003832533210515976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,768,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,768,0.007950933277606964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,512,0.0034122665723164878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,512,0.014434132973353067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,1024,0.008424533406893413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,256,0.002703999976317088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,1536,0.009213866790135701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,256,0.014562132954597472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,128,0.0023999998966852823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,128,0.014553599556287131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,512,0.007469866673151653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,256,0.007145600020885467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,64,0.0021845333278179167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,2048,32,0.002161066730817159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,128,0.006718933085600535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,64,0.014501333236694336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,2048,32,0.013341866930325828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,65536,0.035646935304005936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,16384,0.012558933099110922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,65536,0.034075733025868735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,16384,0.024436267217000325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,12288,0.014199466506640116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,12288,0.02252266605695089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,12288,0.028201599915822346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,10240,0.012498133381207784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,10240,0.020746666193008422
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,16384,0.03509226640065511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,8192,0.011017599701881408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,8192,0.02031360069910685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,10240,0.024807467063268026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,7168,0.010025599598884582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,7168,0.01881386637687683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,8192,0.02067520022392273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,6144,0.009269332885742188
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,6144,0.018451199928919474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,7168,0.019485867023468016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,5120,0.008145066599051159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,5120,0.016713599363962807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,6144,0.0170741339524587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,4096,0.007153066496054332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,4096,0.018601600329081217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,5120,0.01574613352616628
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,3584,0.006534400085608165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,3584,0.016518400112787882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,4096,0.013287466764450074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,3072,0.0055061335364977515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,3072,0.01684160033861796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,3584,0.012470400333404541
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,2560,0.0050784001747767125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,2560,0.01652479966481527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,3072,0.012056533495585125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,2048,0.007109333574771881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,2048,0.016574933131535848
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,2560,0.011212799946467083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,1536,0.005458133419354757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,1536,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,2048,0.009633066256841023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,1024,0.0046304002404212955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,1024,0.014599466323852539
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,2048,65536,0.12350186506907146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,768,0.003576533248027166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,768,0.014575999975204468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,1536,0.009542399644851684
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,512,0.0033418667813142145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,512,0.014141866564750671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,1024,0.008291199803352356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,256,0.002630399912595749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,256,0.014413866400718688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,768,0.007941333452860515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,512,0.007979733248551685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,128,0.0022474666436513263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,128,0.01344000001748403
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,256,0.0070709332823753355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,64,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,128,0.006741333504517872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,64,0.014077867070833841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1536,32,0.002182399978240331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,65536,0.025172267357508344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1536,32,0.013302399714787801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,65536,0.029607466856638592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,16384,0.013678933183352152
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,16384,0.020623999834060668
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,12288,0.010566400488217671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,12288,0.019004799922307334
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,12288,0.02773759961128235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,10240,0.009258666634559631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,10240,0.01869973341623942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,16384,0.03508906761805217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,8192,0.008582400282224019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,8192,0.018617600202560425
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,8192,0.020667733748753865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,7168,0.007531733314196269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,10240,0.024405332406361897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,7168,0.01781439979871114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,6144,0.006725333134333293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,6144,0.0168287992477417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,6144,0.017374932765960693
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,5120,0.00631573349237442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,5120,0.01658560037612915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,7168,0.019014400243759156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,4096,0.005909333129723867
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,4096,0.018778665860493978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,4096,0.013630933562914529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,3584,0.005492266515890757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,3584,0.018630399306615194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,5120,0.015701333681742348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,3072,0.005514666438102722
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,3072,0.016561067104339598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,3584,0.012534399827321371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,3072,0.012053333719571431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,2560,0.004980266590913137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,2560,0.016536532839139303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,2048,0.006233599781990051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,2048,0.016504533092180886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,2560,0.010829866925875346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,2048,0.01001706620057424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,1536,0.005459199845790863
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,1536,0.014723199605941772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,1024,0.004619733492533366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,1024,0.014479999740918478
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,1024,0.008370133241017659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,1536,0.009417600433031718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,768,0.0038272000849246977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,768,0.014545067151387533
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,512,0.003401600072781245
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,512,0.014572800199190775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,768,0.007911466558774312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,256,0.0025888000925381976
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,512,0.0075328002373377485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,256,0.01447466711203257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,128,0.002186666677395503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,128,0.01330880026022593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,256,0.0071263998746871945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,64,0.002198400100072225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,128,0.007076266904671986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,64,0.01444906691710154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,1024,32,0.0021589333812395734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,65536,0.020639999707539877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,1024,32,0.0144405335187912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,65536,0.0321941335995992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,16384,0.011267200112342834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1536,65536,0.12117973168690999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,16384,0.020652800798416138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,12288,0.009633066256841023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,12288,0.018889600038528444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,12288,0.028170667092005414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,10240,0.008508800466855367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,10240,0.0168938676516215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,16384,0.03469760020573934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,8192,0.006983466446399689
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,8192,0.018614399433135986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,10240,0.024919466177622477
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,7168,0.006539733211199443
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,8192,0.020590933163960774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,7168,0.019025067488352455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,6144,0.0068010667959849044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,6144,0.018897066513697304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,7168,0.019006933768590292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,5120,0.0055167997876803074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,5120,0.017691733439763387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,6144,0.017520000537236534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,4096,0.00603413333495458
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,4096,0.01663253307342529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,5120,0.015701333681742348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,3584,0.005502933263778686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,4096,0.013782399892807006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,3584,0.01658986707528432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,3072,0.005515733361244201
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,1024,65536,0.12054399649302165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,3072,0.016295466820398966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,3584,0.012819199760754903
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,2560,0.005115733544031779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,2560,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,3072,0.011611732840538024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,2048,0.004214400053024292
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,2048,0.014565333724021912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,1536,0.005338666836420695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,1536,0.01460693379243215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,2560,0.010930132865905762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,2048,0.009688533345858256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,1024,0.00432533323764801
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,1024,0.014524799585342408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,768,0.0034634667138258614
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,768,0.014595199624697366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,1024,0.008376533786455791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,512,0.003066666672627131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,1536,0.008901333808898926
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,768,0.007623466849327088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,512,0.014517333110173544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,256,0.0026837334036827086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,256,0.014513066411018372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,128,0.0022485333184401194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,512,0.007527466615041096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,256,0.0070720002055168155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,128,0.01392213304837545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,64,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,768,32,0.0021589333812395734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,64,0.01334933340549469
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,128,0.006365866462389629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,65536,0.02680319945017497
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,768,32,0.014164266983668008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,16384,0.00796693315108617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,65536,0.026837333043416338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,16384,0.017517866690953572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,12288,0.007176533341407776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,12288,0.018654932578404747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,12288,0.027833600838979084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,10240,0.006846933563550313
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,10240,0.018296533823013307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,16384,0.03468266725540161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,8192,0.005504000186920166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,8192,0.018911999464035035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,10240,0.024127999941507973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,7168,0.005543466905752818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,8192,0.020594133933385213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,7168,0.01726933320363363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,6144,0.005976533393065134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,6144,0.016681599617004394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,7168,0.0185589333375295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,5120,0.005487999816735586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,5120,0.017434666554133095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,6144,0.017078399658203125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,4096,0.005723733206590017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,4096,0.016698666413625083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,5120,0.01569066643714905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,3584,0.005468800167242686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,3584,0.016500266393025716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,4096,0.013414399822553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,3072,0.00514986664056778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,3072,0.016778665781021117
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,3584,0.012501333157221475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,2560,0.005075199902057648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,2560,0.015282133221626281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,768,65536,0.12034453550974529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,2048,0.004863999783992767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,3072,0.01141973336537679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,2048,0.016100266575813295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,1536,0.004037333279848098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,1536,0.014732799927393594
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,2560,0.010750933488210043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,1024,0.0034175999462604523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,2048,0.010036266843477885
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,1024,0.014471466342608133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,768,0.0030048000315825146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,768,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,1536,0.009179733196894328
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,512,0.00276053324341774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,1024,0.008434133728345235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,512,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,256,0.002678400029738744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,768,0.007500799993673961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,256,0.01448319951693217
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,512,0.007162666817506154
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,128,0.0023093332846959433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,256,0.006692266464233399
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,64,0.002148266633351644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,128,0.013852799932161966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,512,32,0.0022304000953833262
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,64,0.013429333766301474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,128,0.006684799989064534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,65536,0.01694399913152059
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,512,32,0.013425067067146301
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,16384,0.005835733314355215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,65536,0.02250986695289612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,16384,0.017767467101415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,12288,0.005065600077311197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,12288,0.01662720044453939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,12288,0.027591466903686523
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,16384,0.03470933437347412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,10240,0.005559466779232025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,10240,0.017179733514785765
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,10240,0.024430932601292928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,8192,0.005596800148487091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,8192,0.016926934321721397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,7168,0.005393066505591074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,7168,0.017068799336751303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,8192,0.02073600093523661
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,6144,0.005187200009822845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,6144,0.016642133394877114
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,7168,0.019082667430241902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,5120,0.004710400104522705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,5120,0.017002665996551515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,6144,0.017358932892481485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,4096,0.004722133278846741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,4096,0.01718399922053019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,5120,0.015745066603024802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,3584,0.004738133152325948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,3584,0.01653866668542226
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,512,65536,0.12047359943389893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,3072,0.00514986664056778
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,4096,0.013563733299573263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,3072,0.015476266543070475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,2560,0.005203199883302053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,2560,0.014856533209482829
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,3584,0.012862933675448099
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,2048,0.004295466840267182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,3072,0.011271466811498005
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,2048,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,1536,0.0037941334148248037
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,1536,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,2560,0.0114464004834493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,1024,0.0030591999491055804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,2048,0.009959466258684794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,1024,0.014526933431625366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,768,0.003014400104681651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,1536,0.009131733576456707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,768,0.014381866653760275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,512,0.0026079999903837843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,1024,0.008292266726493835
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,768,0.00798933357000351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,512,0.013760000467300415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,256,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,256,0.013926399747530618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,512,0.0070816000302632645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,128,0.002268799891074498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,256,0.006681600213050842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,128,0.014426666498184203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,64,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,256,32,0.002169599880774816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,128,0.007119999825954437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,64,0.014429866274197897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,65536,0.009623466928799947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,256,32,0.013634133338928222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,16384,0.0055189331372578945
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,65536,0.01881813406944275
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,256,65536,0.11952106952667237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,16384,0.018606932957967122
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,12288,0.004725333551565806
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,12288,0.016520532965660095
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,10240,0.005097599824269613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,10240,0.016555733482042947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,12288,0.0272597332795461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,8192,0.005060266455014547
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,16384,0.034629333019256595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,10240,0.024260266621907552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,8192,0.016659200191497803
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,7168,0.004711466530958811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,7168,0.016504533092180886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,6144,0.004648533463478088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,6144,0.016364799936612447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,8192,0.020598399639129638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,7168,0.018589866161346436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,5120,0.004763733347256978
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,5120,0.016711467504501344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,4096,0.004264533519744873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,6144,0.017313067118326822
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,4096,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,3584,0.004632533093293508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,3584,0.015608533223470052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,5120,0.01529813309510549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,3072,0.004317866762479147
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,4096,0.013293866316477457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,3072,0.01662826637427012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,2560,0.0046528001626332605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,3584,0.012472533186276754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,2560,0.015990400314331056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,2048,0.004294399917125702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,3072,0.0116565336783727
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,2048,0.015007999539375306
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,1536,0.0038453333079814913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,2560,0.01109333336353302
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,1536,0.014596266547838846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,2048,0.009958400328954061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,1024,0.003099733342727025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,1024,0.013252266248067222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,768,0.002605866640806198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,1536,0.009146666526794434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,1024,0.008035199840863545
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,768,0.013179733355840047
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,512,0.0022495999932289124
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,768,0.007896533111731212
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,512,0.012761599818865457
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,256,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,256,0.012653866410255432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,512,0.007238399982452392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,128,0.0022111999491850535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,256,0.006685866912206013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,128,0.013169067104657492
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,128,0.006286933521429698
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,64,0.0018911999960740406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,64,0.0127402663230896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,128,32,0.0018687999496857326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,128,32,0.012781866391499839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,65536,0.006807466844717662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,16384,0.005899733304977417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,65536,0.01885653336842855
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,12288,0.005020800232887268
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,12288,0.01662720044453939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,16384,0.018106667200724284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,10240,0.004689066608746847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,8192,0.005008000135421753
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,10240,0.01655893325805664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,8192,0.016523733735084534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,7168,0.004646400113900503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,6144,0.004647466540336609
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,2,128,65536,0.11991999944051106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,7168,0.016540799538294473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,5120,0.004251733422279358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,4096,0.004667733112970988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,5120,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,6144,0.016571733355522155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,3584,0.004622933268547058
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,4096,0.016578132907549538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,3072,0.00421973317861557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,3584,0.015093333522478738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,2560,0.004613333443800608
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,2048,0.004222933451334635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,3072,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,2560,0.014522666732470194
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,1536,0.0038389332592487337
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,2048,0.014454399545987448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,1024,0.003053866575161616
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,1536,0.01455466647942861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,768,0.0030421334008375804
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,1024,0.013724799950917563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,512,0.0026261332134405774
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,256,0.002203733225663503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,768,0.013680000106493631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,128,0.0022229333718617756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,512,0.012991999586423239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,64,0.0018207999567190808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,256,0.01383573313554128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,128,0.01275200049082438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,64,32,0.0021909333765506743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,64,0.012611200412114462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,65536,0.006409599880377452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,16384,0.005456000069777171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,64,32,0.01251520017782847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,65536,0.018611200650533042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,12288,0.005109333495299021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,10240,0.005044266581535339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,16384,0.017117865880330405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,12288,0.016505600015322367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,8192,0.0046517332394917805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,7168,0.00505920002857844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,10240,0.016313599546750386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,8192,0.016504533092180886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,6144,0.005110399921735128
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,5120,0.00420906643072764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,6144,0.016166399916013083
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,7168,0.016824533541997276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,4096,0.003866666555404663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,5120,0.01651306649049123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,3072,0.004644266764322917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,3584,0.004308266441027323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,4096,0.01648640036582947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,3584,0.014586666226387024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,2560,0.004340266684691111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,2048,0.004208000004291534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,3072,0.01461013356844584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,2560,0.014490666985511779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,1536,0.0033952000240484873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,1024,0.0029781334102153777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,1536,0.014509866635004679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,2048,0.014698666334152222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,768,0.0025781333446502686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,512,0.0025301332275072734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,1024,0.013790933291117349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,256,0.002194133400917053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,768,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,512,0.012664533654848733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,128,0.0021717332303524016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,64,0.0017557332913080853
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,256,0.012570666273434958
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,128,0.01358080009619395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,2,32,32,0.0018005333840847015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,64,0.014064000050226847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,2,32,32,0.013071999947230021
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,16384,0.17391573588053386
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,16384,0.3262869199117025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,12288,0.23348479270935057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,12288,0.1381685256958008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,10240,0.19829546610514323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,10240,0.11682026386260987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,8192,0.16002240180969238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,8192,0.10115626653035481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,8192,0.07065813541412354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,7168,0.1395359992980957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,7168,0.08332800070444743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,10240,0.07863146464029948
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,6144,0.12327146530151367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,6144,0.07350293000539145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,12288,0.10296213626861572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,5120,0.10221439997355145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,5120,0.06391040086746216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,6144,0.05367679993311564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,4096,0.08721919854482016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,4096,0.053523198763529456
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,7168,0.06239786545435587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,3584,0.0703765312830607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,3584,0.04937386512756348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,5120,0.047585066159566244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,3072,0.06099840005238851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,3072,0.04503786563873291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,16384,0.12862506707509358
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,2560,0.05125759840011597
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,2560,0.04038293361663818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,3584,0.03550399939219157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,2048,0.041702401638031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,2048,0.03495573202768962
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,4096,0.03863573471705119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,1536,0.03266666730244954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,1536,0.028963200251261395
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,2560,0.026461867491404216
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,1024,0.022241065899531044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,1024,0.024676267306009927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,3072,0.03184000054995219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,768,0.018011732896169027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,768,0.022210133075714112
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,2048,0.023652267456054688
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,512,0.012459733088811238
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,512,0.01934400002161662
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,1536,0.01952000061670939
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,256,0.007758933305740357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,256,0.01663253307342529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,1024,0.015570132931073507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,128,0.004826666911443075
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,128,0.016586666305859886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,128,0.008358400066693623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,64,0.00421973317861557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,64,0.016476800044377647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,65536,32,0.004203733305136363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,65536,32,0.01646613379319509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,768,0.013989333311716715
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,65536,0.31932268142700193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,65536,0.17666133244832355
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,16384,0.09035627047220865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,16384,0.05989120006561279
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,256,0.009562666217486065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,12288,0.0656544009844462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,12288,0.043339733282725015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,65536,512,0.011236266295115153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,10240,0.055447467168172206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,10240,0.03914453188578288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,10240,0.03469119866689046
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,8192,0.04642239809036255
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,8192,0.03439146677652995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,12288,0.03709973494211833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,7168,0.035922133922576906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,7168,0.03175680041313171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,16384,0.051164801915486655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,6144,0.03128213286399841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,6144,0.029104000329971312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,8192,0.028967465957005816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,5120,0.026789333422978717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,5120,0.026794666051864625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,7168,0.025205334027608235
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,4096,0.021995733181635536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,4096,0.024804266293843587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,5120,0.020441599686940513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,3584,0.019458132982254028
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,3584,0.022696532805760703
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,6144,0.023076266050338745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,3072,0.017054933309555053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,3072,0.021223467588424683
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,4096,0.01737706661224365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,2560,0.014587733149528503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,2560,0.020731733242670695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,2560,0.013529599706331889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,2048,0.01243946651617686
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,2048,0.018872533241907755
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,3584,0.016157866517702738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,1536,0.010088533163070679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,1536,0.01848213275273641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,3072,0.014882133404413859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,1024,0.007663999994595845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,1024,0.01456106702486674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,1024,0.008774399757385254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,768,0.006424533327420552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,1536,0.010035199920336406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,768,0.01476800044377645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,512,0.004108799993991852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,512,0.014830933014551798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,2048,0.011546666423479717
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,256,0.003081600119670232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,256,0.014145066340764364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,512,0.007930666704972585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,128,0.0026378666361172995
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,128,0.013723733027776084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,256,0.007623466849327088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,64,0.002233600119749705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,64,0.01383039951324463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,16384,32,0.002274133265018463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,16384,32,0.013272533814112345
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,128,0.007062399884064992
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,65536,0.23109332720438638
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,65536,0.13786880175272626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,768,0.008354133367538452
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,16384,0.06855466365814208
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,16384,0.04844586849212647
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,12288,0.049029334386189775
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,12288,0.041555198033650716
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,12288,0.02992853323618571
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,10240,0.04166826804478963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,10240,0.03753173351287842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,16384,0.03825386762619019
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,8192,0.03484586477279663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,8192,0.03283413251241048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,10240,0.0256330668926239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,7168,0.03056959907213847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,7168,0.03178986708323161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,8192,0.02202026645342509
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,6144,0.027505065997441607
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,6144,0.028649600346883138
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,7168,0.02039253314336141
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,5120,0.02448106606801351
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,5120,0.02696746587753296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,6144,0.018206934134165444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,4096,0.020337067047754922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,4096,0.02474666635195414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,5120,0.016665599743525186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,3584,0.018181333939234413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,3584,0.020660267273585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,4096,0.014638933539390563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,3072,0.015953066945075988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,3072,0.02060479919115702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,3584,0.013452800114949546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,2560,0.014201600352923074
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,2560,0.018605866034825645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,3072,0.012174933155377706
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,2048,0.012427733341852824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,2048,0.0186954657236735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,2560,0.01088746686776479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,1536,0.008360532919565837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,1536,0.015847466389338174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,1536,0.009153067072232565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,1024,0.006342400113741558
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,1024,0.014504533012708029
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,2048,0.010102400183677673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,768,0.004340266684691111
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,768,0.014587733149528503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,1024,0.008364799618721008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,512,0.0035189333061377203
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,512,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,768,0.008082133531570435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,256,0.002690133452415466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,256,0.01456000010172526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,512,0.007505066692829132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,128,0.0025920001169045764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,128,0.012963199615478515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,128,0.006681600213050842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,64,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,256,0.0069909334182739254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,64,0.013748266299565635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,12288,32,0.002239999920129776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,12288,32,0.013859200477600097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,65536,0.19343360265096027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,65536,0.11885759830474854
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,16384,0.05830719868342081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,16384,0.04384746551513672
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,16384,0.036695468425750735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,12288,0.04103999932607015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,12288,0.037109335263570145
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,12288,0.028921600182851153
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,10240,0.035396265983581546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,10240,0.03360746701558431
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,16384,65536,0.1643199920654297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,8192,0.028717867533365887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,8192,0.031036800146102904
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,10240,0.0255840003490448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,7168,0.02755519946416219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,7168,0.029076266288757324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,8192,0.021840000152587892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,6144,0.023847466707229613
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,6144,0.028566400210062664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,7168,0.020261333386103312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,5120,0.020731733242670695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,5120,0.026428800821304322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,12288,65536,0.1324405352274577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,4096,0.017189333836237587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,4096,0.022683733701705934
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,6144,0.018538665771484376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,3584,0.016425599654515587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,3584,0.021859200795491536
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,3584,0.013255467017491659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,3072,0.014666666587193808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,4096,0.01458026667435964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,3072,0.020645334323247274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,2560,0.01323199967543284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,2560,0.02002560098965963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,5120,0.016693333784739174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,2048,0.010105599959691364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,2048,0.018465065956115724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,2048,0.009973333279291789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,1536,0.00876586635907491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,1536,0.016570666432380678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,2560,0.010890666643778484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,1024,0.006457599997520447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,1024,0.01656426688035329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,3072,0.01225493351618449
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,768,0.005239466826121012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,768,0.014652799566586813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,1024,0.008132266501585644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,512,0.004305066665013632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,512,0.014659200112024942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,1536,0.009141332904497783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,256,0.002605866640806198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,256,0.013809067010879517
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,256,0.006726400057474773
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,128,0.0022026665508747103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,128,0.01381653348604838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,768,0.008041599889596303
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,64,0.002254933367172877
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,64,0.01381013294061025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,10240,32,0.0023146666586399078
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,128,0.006727466483910878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,10240,32,0.013550933202107748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,65536,0.15420799255371093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,65536,0.09690133730570474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,512,0.007526400188604991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,16384,0.04760640064875285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,16384,0.03612266778945923
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,12288,0.03502293427785237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,12288,0.03062506715456645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,12288,0.028807467222213744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,10240,0.030070400238037108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,10240,0.02868480086326599
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,16384,0.03626559972763062
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,8192,0.022798933585484824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,8192,0.025231999158859254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,10240,0.025165865818659466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,7168,0.019910399119059244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,7168,0.024289067586263022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,8192,0.02165973385175069
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,6144,0.018052266041437785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,6144,0.022745599349339805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,7168,0.020011732975641884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,5120,0.015094400445620219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,5120,0.021268266439437866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,6144,0.01835839947064718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,4096,0.01297920048236847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,4096,0.02035520076751709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,10240,65536,0.1272543986638387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,3584,0.011667199929555257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,3584,0.018757333358128868
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,5120,0.0164000004529953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,3072,0.010248532891273499
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,3072,0.018573866287867228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,3584,0.013268267114957174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,2560,0.008917333682378133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,2560,0.0165994664033254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,4096,0.014075733224550881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,2048,0.007784533500671387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,2048,0.01609386702378591
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,3072,0.01200320025285085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,1536,0.006303999821345012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,1536,0.014711466431617738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,2560,0.010811733206113179
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,1024,0.004247466723124186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,1024,0.01446293294429779
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,2048,0.010053333640098572
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,768,0.0038015998899936674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,768,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,1536,0.009585066636403402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,512,0.003387733300526937
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,512,0.014425599575042724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,1024,0.008342400193214417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,256,0.0025973332424958544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,256,0.013521066308021546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,256,0.007073066631952922
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,128,0.002218666672706604
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,128,0.013633066415786743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,768,0.007945600152015685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,64,0.0021856000026067098
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,64,0.013646933436393737
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,8192,32,0.0021674667795499166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,512,0.007520000139872233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,8192,32,0.013264000415802002
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,128,0.006704000135262807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,65536,0.1361621379852295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,65536,0.08593493302663167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,16384,0.040491731961568196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,16384,0.03603519996007283
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,12288,0.03148373365402222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,12288,0.028489599625269573
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,12288,0.028683733940124512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,10240,0.02683839996655782
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,10240,0.026472532749176027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,16384,0.03584959904352824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,8192,0.022291199366251627
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,8192,0.024055467049280802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,10240,0.02513386607170105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,7168,0.01962560017903646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,7168,0.02271040081977844
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,8192,0.021710934241612752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,6144,0.01585919956366221
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,6144,0.021146667003631592
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,7168,0.019834667444229126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,5120,0.014077867070833841
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,5120,0.0206058661142985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,6144,0.018202666441599527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,4096,0.011565867066383361
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,4096,0.018703999121983846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,5120,0.016241066654523215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,3584,0.010744532942771912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,3584,0.018593066930770875
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,4096,0.013672533631324767
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,3072,0.009302399555842082
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,3072,0.0167413334051768
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,3584,0.012831999858220419
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,2560,0.00846613347530365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,2560,0.016557866334915163
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,3072,0.012075733145078022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,2048,0.007161599894364675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,2048,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,2560,0.011244799693425496
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,1536,0.005621333420276642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,1536,0.014535466829935709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,2048,0.009967999656995137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,1024,0.004223999877770742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,1024,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,1536,0.009595732887585957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,768,0.003802666564782461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,768,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,768,0.007922133306662242
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,512,0.0030559999247392017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,512,0.014327466487884521
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,1024,0.008171733220418293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,256,0.002616533388694127
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,256,0.014045866330464682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,256,0.007089066505432129
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,128,0.002182399978240331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,128,0.013642666737238565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,512,0.007500799993673961
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,64,0.0021888000269730886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,64,0.013665067156155905
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,7168,32,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,7168,32,0.01362879971663157
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,128,0.006664533416430156
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,65536,0.11834666728973389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,65536,0.0790165344874064
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,16384,0.03741653362909953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,16384,0.03390613396962484
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,8192,65536,0.12553919951121012
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,12288,0.02754240036010742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,12288,0.028812799851099653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,16384,0.03582186698913574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,10240,0.024477867285410564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,10240,0.02722559968630473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,12288,0.028888533512751263
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,8192,0.01954560081164042
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,8192,0.02273599902788798
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,8192,0.021512534221013388
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,7168,0.017435733477274576
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,7168,0.021556266148885093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,10240,0.025230934222539265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,6144,0.01643519997596741
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,6144,0.020653865734736123
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,7168,0.01969919999440511
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,5120,0.013774933417638144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,5120,0.019792000452677407
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,6144,0.017864533265431724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,4096,0.01065066655476888
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,4096,0.018643200397491455
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,5120,0.01574399967988332
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,3584,0.00970240036646525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,3584,0.017192532618840538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,4096,0.013861333330472311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,3072,0.008614400029182434
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,3072,0.016673066218694053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,3072,0.012177067001660664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,2560,0.00765119989713033
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,2560,0.01511253317197164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,3584,0.01285866697629293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,2048,0.006757333377997081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,2048,0.014539733529090881
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,7168,65536,0.1252031962076823
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,1536,0.005207466582457224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,2048,0.009962667028109233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,1536,0.014460800091425577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,1024,0.0046293333172798155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,1024,0.01493013302485148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,2560,0.011282133062680562
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,768,0.0036256000399589538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,768,0.014582399527231851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,768,0.007982933521270752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,512,0.0031008000175158186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,512,0.014086400469144186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,1024,0.008424533406893413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,256,0.0026677332818508146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,1536,0.009166933099428813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,256,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,128,0.002311466634273529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,128,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,128,0.00680213322242101
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,64,0.0022581333915392557
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,256,0.0070613334576288865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,64,0.01418346663316091
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,6144,32,0.002164266755183538
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,6144,32,0.014587733149528503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,65536,0.10009280045827229
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,512,0.007656533519426982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,65536,0.06941013336181641
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,16384,0.03201280037562053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,16384,0.03181653420130412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,12288,0.02403093377749125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,12288,0.027049599091211955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,12288,0.02847359975179036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,10240,0.02050986687342326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,10240,0.025870933135350542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,16384,0.035811201731363936
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,8192,0.017935999234517417
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,8192,0.021197867393493653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,10240,0.025154133637746174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,7168,0.01576533317565918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,7168,0.020811732610066733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,8192,0.021517866849899293
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,6144,0.01416853368282318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,6144,0.020336000124613444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,7168,0.019479467471440633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,5120,0.012458667159080505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,5120,0.019195733467737834
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,6144,0.017359999815622966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,4096,0.009782399733861287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,4096,0.017573332786560057
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,5120,0.015827199816703795
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,3584,0.008742400010426839
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,3584,0.016586666305859886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,4096,0.013309866189956665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,3072,0.007737599809964498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,3072,0.016115199526151022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,3584,0.012876799702644348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,2560,0.007141333321730297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,2560,0.015432533621788026
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,3072,0.011619200309117634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,2048,0.005857066810131073
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,2048,0.014477866888046264
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,2560,0.010819199681282043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,1536,0.004975999891757965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,1536,0.01456106702486674
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,2048,0.00965226689974467
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,1024,0.004003199934959412
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,1024,0.014519466956456503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,1536,0.009219200412432352
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,768,0.003455999990304311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,6144,65536,0.12555627028147381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,768,0.014484266440073649
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,512,0.0029546665648619336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,512,0.014436266819636025
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,512,0.007472000022729237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,256,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,256,0.014443733294804893
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,768,0.007962666451931
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,128,0.0021664001047611236
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,128,0.013371733824412027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,256,0.007117866476376851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,64,0.002199466774861018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,1024,0.008344533046086629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,64,0.01386666695276896
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,5120,32,0.0021568000316619872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,128,0.006771199901898702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,5120,32,0.014416000247001648
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,65536,0.08163839975992838
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,65536,0.05858773390452067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,16384,0.025690666834513348
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,16384,0.02699199914932251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,12288,0.02053440014521281
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,12288,0.024987733364105223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,12288,0.028832000494003297
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,10240,0.017650133371353148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,10240,0.02304853399594625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,16384,0.03583360115687052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,8192,0.01458346645037333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,8192,0.02258239984512329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,10240,0.02513386607170105
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,7168,0.01360106666882833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,7168,0.023883734146753946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,8192,0.021130667130152384
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,6144,0.012185600399971009
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,6144,0.022616533438364665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,7168,0.019453867276509603
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,5120,0.011090133587519329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,5120,0.021590399742126464
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,6144,0.017359999815622966
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,4096,0.009296000003814697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,4096,0.018603734175364175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,5120,0.01575573285420736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,3584,0.012006400028864543
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,3584,0.018600533405939736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,4096,0.013661866386731466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,3072,0.010947199662526448
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,3072,0.018257067600886027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,3584,0.012478933731714884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,2560,0.009607467055320739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,2560,0.016583466529846193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,2560,0.011243733763694762
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,2048,0.008229333162307739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,2048,0.016551466782887776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,3072,0.012040533622105916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,1536,0.006312533219655354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,1536,0.015335466464360556
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,1536,0.00914026697476705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,1024,0.005037866532802582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,1024,0.014532267053922018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,2048,0.010001066327095031
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,768,0.004221866528193155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,768,0.014460800091425577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,768,0.007971199850241344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,512,0.003457066665093104
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,512,0.014257066448529563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,1024,0.007906133433183034
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,256,0.0026176000634829206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,256,0.013317333658536276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,256,0.007100800176461537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,128,0.002178133279085159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,128,0.012955733140309653
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,512,0.007898666461308797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,64,0.0021674667795499166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,128,0.006685866912206013
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,64,0.013271466890970866
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,4096,32,0.0021503999829292296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,4096,32,0.013527466853459676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,65536,0.07281920115152994
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,65536,0.054231464862823486
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,16384,0.021988266706466676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,16384,0.02579306761423747
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,5120,65536,0.12509120305379232
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,12288,0.017339734236399333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,12288,0.02280319929122925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,12288,0.02845226724942525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,10240,0.01525973379611969
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,10240,0.022830933332443237
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,16384,0.035897600650787356
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,8192,0.012928000092506409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,8192,0.022091732422510783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,10240,0.02486293315887451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,7168,0.012478933731714884
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,7168,0.022652800877889
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,8192,0.021269333362579346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,6144,0.011890133221944172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,6144,0.021847466627756752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,7168,0.019102933009465535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,5120,0.00999679962793986
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,5120,0.019818667570749918
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,6144,0.017874133586883546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,4096,0.009227733810742695
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,4096,0.018707199891408285
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,4096,0.013843199610710144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,3584,0.011080533266067505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,3584,0.018705066045125326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,5120,0.015703466534614564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,3072,0.010874666770299276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,3072,0.016563199957211814
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,3072,0.01202133297920227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,2560,0.009567999839782714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,2560,0.016497066617012023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,3584,0.012615467111269632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,2048,0.008088533580303193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,2048,0.01669973333676656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,4096,65536,0.12492480278015136
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,1536,0.006832000116507213
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,1536,0.01613759994506836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,2560,0.010869333148002624
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,1024,0.005036800106366476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,1024,0.014564266800880432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,2048,0.010088533163070679
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,768,0.004392533500989278
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,768,0.014631467064221701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,1536,0.009199999769528707
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,512,0.0035690667728583017
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,512,0.014550399780273438
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,1024,0.007926400005817413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,256,0.00257493332028389
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,256,0.013727999726931252
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,512,0.007657599945863088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,128,0.002324266731739044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,768,0.007997866471608479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,128,0.014457600315411887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,128,0.006798933446407318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,256,0.007108266651630402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,64,0.002299733211596807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,64,0.013451733191808066
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3584,32,0.002311466634273529
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3584,32,0.013011200229326883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,65536,0.06327999830245971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,65536,0.04914559920628865
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,16384,0.020177066326141357
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,16384,0.02522346576054891
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,12288,0.016114133596420287
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,12288,0.023423999547958374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,12288,0.02844160000483195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,10240,0.014033066232999167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,10240,0.023170133431752525
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,16384,0.03551359971364339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,8192,0.012290133039156596
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,8192,0.020722132921218873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,10240,0.02487893303235372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,7168,0.011136000355084736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,7168,0.022855466604232787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,8192,0.02067520022392273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,6144,0.010427733262379963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,6144,0.020660267273585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,7168,0.019128533204396565
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,6144,0.01736746629079183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,5120,0.009454933802286784
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,5120,0.01869973341623942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,4096,0.007994666695594788
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,4096,0.018374399344126383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,4096,0.013274666666984559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,3584,0.010770133137702942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,3584,0.018583466609319053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,5120,0.015876266360282897
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,3072,0.009283199906349182
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,3072,0.01865066687266032
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,3584,0.01279146671295166
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,2560,0.008220799763997396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,2560,0.016598400473594666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,3072,0.011798399686813354
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,2048,0.007784533500671387
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,2048,0.01676693360010783
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3584,65536,0.12534399827321369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,1536,0.006516266862551372
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,1536,0.015904000401496886
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,2560,0.010822400450706482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,1024,0.005074133475621542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,1024,0.01452906628449758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,2048,0.009754666686058044
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,768,0.004422399898370107
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,768,0.014707199732462563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,1536,0.00913279950618744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,512,0.003538133452335993
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,1024,0.008053333560625712
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,512,0.01461013356844584
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,256,0.0026848000784715016
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,256,0.01453013320763906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,768,0.007939200103282928
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,256,0.007097599903742473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,128,0.002551466723283132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,128,0.013296000162760415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,512,0.0074976002176602675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,64,0.0021685334543387097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,3072,32,0.0022069332500298817
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,64,0.013033599654833476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,3072,32,0.013750400145848593
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,65536,0.055821867783864346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,128,0.006716800232728322
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,16384,0.018026665846506754
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,65536,0.044232531388600664
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,16384,0.024451200167338052
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,12288,0.014173866311709086
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,12288,0.022490666309992472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,12288,0.028100266059239702
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,10240,0.01248426636060079
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,10240,0.021691733598709108
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,16384,0.03560426632563273
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,8192,0.010913067062695821
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,8192,0.020503467321395873
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,10240,0.024425599972407022
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,7168,0.010009599725405376
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,7168,0.019012266397476198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,8192,0.020668800671895346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,6144,0.00913279950618744
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,6144,0.020430932442347206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,7168,0.018910932540893554
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,5120,0.008457600076993307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,5120,0.01873706579208374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,6144,0.017347200711568197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,4096,0.007178666690985362
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,4096,0.018259199460347493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,5120,0.01575573285420736
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,3584,0.006345599889755249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,3584,0.017951999107996622
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,4096,0.013306666413942972
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,3072,0.007963733375072479
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,3072,0.016522666811943053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,3584,0.012845866878827415
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,2560,0.007311999797821045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,2560,0.016569599509239197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,3072,0.011687466502189636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,2048,0.007243733108043671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,2048,0.016639999548594155
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,3072,65536,0.12556053002675374
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,1536,0.006257066627343495
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,1536,0.015100799997647605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,2560,0.010821333527565003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,1024,0.005092266698678335
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,1024,0.014516266187032065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,2048,0.01001706620057424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,768,0.004233600199222564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,768,0.01460906664530436
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,1024,0.008177066842714947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,512,0.0035135999321937563
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,512,0.014481066664059957
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,1536,0.00921493371327718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,256,0.002630399912595749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,256,0.014497066537539164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,512,0.007080533107121785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,128,0.0025802666942278544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,256,0.007506133119265239
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,128,0.013582932949066161
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,64,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,768,0.007892266909281413
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2560,32,0.0021322667598724365
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,64,0.013820800185203552
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2560,32,0.01442346672217051
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,65536,0.04610986709594726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,128,0.006677333513895671
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,65536,0.04028373161951701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,16384,0.015362133582433065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,16384,0.022653865814208984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,12288,0.011639466881752015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,12288,0.02062826752662659
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,12288,0.027730133136113482
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,10240,0.010402133067448933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,10240,0.020296533902486168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,16384,0.035444267590840656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,8192,0.009212799866994222
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,8192,0.020644267400105797
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,10240,0.024489599466323852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,7168,0.008833066622416178
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,7168,0.020773333311080933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,8192,0.021078399817148843
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,6144,0.007931733131408691
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,6144,0.01862293283144633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,7168,0.018979199727376304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,5120,0.006757333377997081
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,5120,0.018633600076039633
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,6144,0.017802667617797852
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,4096,0.009240532914797466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,4096,0.018651733795801796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,5120,0.015706666310628257
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,3584,0.008017066617806752
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,3584,0.01853333314259847
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,4096,0.013731200496355692
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,3072,0.006679466863473256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,3072,0.016952532529830932
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,3584,0.012805333733558655
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,2560,0.00586346685886383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,2560,0.01665173371632894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,3072,0.012029866377512615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,2048,0.007515733440717061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,2048,0.016497066617012023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,2560,0.01114453375339508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,1536,0.006263466676076253
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,1536,0.01552959978580475
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,2048,0.010386133193969726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,1024,0.0046410664916038515
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,1024,0.014514133334159851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2560,65536,0.12445440292358398
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,768,0.0038751999537150065
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,768,0.014526933431625366
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,1024,0.00832533339659373
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,512,0.0034111998975276947
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,1536,0.00920746624469757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,512,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,256,0.002605866640806198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,256,0.014065066973368326
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,768,0.007899733384450276
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,128,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,256,0.007512533167997997
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,512,0.007656533519426982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,128,0.013327999909718832
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,64,0.002250666668017705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,2048,32,0.0026026666164398193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,64,0.013754666845003764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,65536,0.035403732458750406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,2048,32,0.014528000354766845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,128,0.006806399921576183
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,16384,0.012471466263135274
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,65536,0.03395093282063802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,16384,0.02485333283742269
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,12288,0.014520532886187234
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,12288,0.022027732928593953
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,12288,0.0281877338886261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,10240,0.013083733121554055
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,10240,0.020872533321380615
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,16384,0.03511999845504761
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,8192,0.0120170662800471
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,8192,0.020806399981180827
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,10240,0.024965333938598632
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,7168,0.010658133029937743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,7168,0.018838399648666383
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,8192,0.020662399133046468
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,6144,0.010002133250236512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,6144,0.018268799781799315
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,7168,0.019197867314020792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,5120,0.008520533641179402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,5120,0.016736000776290894
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,6144,0.01692906618118286
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,4096,0.007550933460394542
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,4096,0.018066134055455527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,5120,0.015892266233762106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,3584,0.006837333242098491
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,3584,0.016684800386428833
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,4096,0.013288533687591553
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,3072,0.005883733431498209
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,3072,0.016541866461435954
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,3584,0.012995200355847678
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,2560,0.005209599932034811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,2560,0.016645333170890807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,3072,0.011590400338172912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,2048,0.006538666784763336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,2048,0.01474133332570394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,2560,0.010909866293271382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,1536,0.005949866771697998
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,1536,0.014565333724021912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,2048,0.009553066889444987
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,1024,0.004103466620047887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,1024,0.014525866508483887
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,2048,65536,0.12344746589660645
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,1536,0.009190400441487629
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,768,0.0040554667512575785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,768,0.014582399527231851
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,512,0.00311253344019254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,512,0.01451520025730133
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,1024,0.008327466746171314
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,256,0.0025770666698614756
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,256,0.014004266262054444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,256,0.006679466863473256
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,768,0.008071466783682505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,128,0.002216533323129018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,512,0.007607466479142506
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,64,0.002232533444960912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,128,0.013422933220863343
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1536,32,0.0023007998863856
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,64,0.013585066795349121
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,128,0.0062837332487106325
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1536,32,0.013667200009028116
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,65536,0.025014400482177734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,65536,0.027116799354553224
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,16384,0.013904000322024027
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,16384,0.021780266364415487
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,12288,0.011158399780591329
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,12288,0.018663465976715088
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,12288,0.027590399980545043
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,10240,0.009398399790128072
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,10240,0.01902186671892802
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,16384,0.03540906508763631
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,8192,0.009001599748929341
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,8192,0.01880106727282206
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,10240,0.023891200621922813
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,7168,0.00817386656999588
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,7168,0.018389334281285606
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,8192,0.020850133895874024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,6144,0.007231999933719635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,6144,0.01689493258794149
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,7168,0.01904426614443461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,5120,0.0067221333583196
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,5120,0.016546133160591125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,6144,0.017518933614095053
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,4096,0.006374399860699971
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,4096,0.018760534127553304
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,5120,0.015260799725850423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,3584,0.005443199972311656
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,3584,0.01752426624298096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,3584,0.01244160036245982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,3072,0.005463466544946035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,3072,0.016707199811935424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,4096,0.013782399892807006
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,2560,0.005176533261934916
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,2560,0.016636799772580466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,3072,0.011589333415031433
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,2048,0.007158400118350982
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,2560,0.011313066879908244
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,2048,0.016157866517702738
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,1536,0.0059349333246548975
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,1536,0.014532267053922018
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,1536,0.00909440020720164
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,1024,0.004590933521588644
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,1024,0.014467199643452963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,2048,0.010107733805974324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,768,0.003952000041802724
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,768,0.014633599917093912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,1024,0.008269866804281871
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,512,0.0034314667185147605
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,512,0.014435199896494546
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,768,0.008036266764005024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,256,0.002700799951950709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,512,0.007155199845631917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,256,0.01411626636981964
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,256,0.00665280024210612
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,128,0.0022143999735514323
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,128,0.014083199699719749
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,128,0.006252799928188324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,64,0.002201599876085917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,64,0.013713066776593527
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,1024,32,0.0022079999248186748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,1024,32,0.01405333379904429
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,65536,0.020616533358891805
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1536,65536,0.12117546399434406
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,65536,0.030932267506917317
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,16384,0.011347200473149617
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,16384,0.019338667392730713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,12288,0.010479999581972758
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,12288,0.01950826644897461
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,16384,0.03455573320388794
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,10240,0.008098133405049642
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,10240,0.017590399583180746
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,12288,0.028369067112604777
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,8192,0.007277866701285045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,8192,0.01869759956995646
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,8192,0.02117439905802409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,7168,0.006775466601053874
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,10240,0.024116265773773193
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,7168,0.017553067207336424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,6144,0.00680320014556249
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,6144,0.016768000523249307
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,7168,0.018986666202545167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,5120,0.005448533097902933
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,5120,0.01705066760381063
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,6144,0.01747093399365743
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,4096,0.0059456000725428265
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,4096,0.016565333803494772
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,1024,65536,0.12053226629892985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,5120,0.015387733777364096
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,3584,0.0060693333546320595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,4096,0.013771733641624451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,3584,0.016568533579508462
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,3072,0.005481599768002828
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,3072,0.01662613352139791
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,2560,0.0050794666012128195
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,2560,0.01632426679134369
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,3072,0.01156160036722819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,2048,0.0042346666256586705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,3584,0.012504532933235168
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,2560,0.010835199554761251
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,2048,0.014677332838376364
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,1536,0.0055402666330337524
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,1536,0.014433067043622336
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,1024,0.004154666761557261
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,1024,0.014505599935849508
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,1536,0.00920746624469757
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,1024,0.0079434668024381
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,768,0.003822933385769526
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,768,0.014456533392270408
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,512,0.002996266633272171
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,2048,0.009995733698209126
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,512,0.014124799768129984
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,256,0.0025653332471847535
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,256,0.014070399602254233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,768,0.008040533463160197
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,512,0.007609599828720092
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,128,0.002253866692384084
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,256,0.006727466483910878
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,64,0.002165333429972331
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,128,0.013314132889111837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,768,32,0.002279466638962428
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,128,0.0067093332608540845
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,64,0.013729066650072733
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,65536,0.02820693254470825
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,768,32,0.014167466759681701
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,65536,0.026121600468953447
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,16384,0.008037333190441132
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,16384,0.018329600493113198
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,12288,0.007514666517575581
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,12288,0.018565332889556883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,12288,0.027276800076166792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,10240,0.007111466427644093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,10240,0.01873813271522522
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,16384,0.03470613161722819
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,8192,0.005613866448402405
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,8192,0.018628267447153728
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,10240,0.023922133445739745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,7168,0.005868799984455109
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,7168,0.018041600783665977
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,8192,0.020807466904322305
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,6144,0.006004266440868378
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,6144,0.01842026710510254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,7168,0.018956800301869713
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,5120,0.005496533215045929
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,5120,0.01685333251953125
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,6144,0.017458132902781167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,4096,0.005505066613356272
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,4096,0.016635732849438985
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,5120,0.015356799960136414
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,3584,0.005070933202902476
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,3584,0.015875200430552162
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,4096,0.013358933726946512
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,3584,0.01246506671110789
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,3072,0.005031466484069824
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,3072,0.01513706644376119
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,3072,0.011577600240707397
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,2560,0.004680533210436503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,2560,0.014459733168284097
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,768,65536,0.1202997366587321
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,2048,0.00494400014479955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,2048,0.015094400445620219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,2560,0.010770133137702942
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,1536,0.003864533454179764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,1536,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,2048,0.010044800241788228
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,1024,0.003433600068092346
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,1536,0.009174399574597676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,768,0.002977066735426585
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,1024,0.01511360009511312
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,768,0.014393599828084311
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,512,0.0026357332865397137
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,512,0.01404906709988912
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,1024,0.008019199967384339
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,256,0.0025973332424958544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,768,0.008116266628106435
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,256,0.014476799964904785
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,512,0.007075199981530507
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,128,0.002234666546185811
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,128,0.013914666573206582
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,64,0.002195200075705846
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,256,0.007207466661930085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,128,0.006285866598288219
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,512,32,0.0021589333812395734
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,64,0.013435733318328858
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,512,32,0.01446613371372223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,65536,0.01574720044930776
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,16384,0.005485866467158
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,65536,0.02085226575533549
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,16384,0.016596266627311708
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,12288,0.005038933455944061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,12288,0.018915200233459474
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,12288,0.0276853342851003
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,10240,0.005544533332188925
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,16384,0.034499200185139973
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,10240,0.016638933618863424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,8192,0.0051701332132021586
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,8192,0.016595199704170227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,10240,0.024345600605010988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,7168,0.0050901333491007485
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,8192,0.020680532852808634
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,7168,0.016772266228993735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,6144,0.004740266501903534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,6144,0.016518400112787882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,6144,0.01729493339856466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,7168,0.01906026601791382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,5120,0.004712533454100291
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,5120,0.01660160024960836
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,4096,0.004315733412901561
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,4096,0.016618667046229045
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,5120,0.015271466970443726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,3584,0.004667733112970988
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,3584,0.015390933553377787
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,4096,0.01409066617488861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,3072,0.004644266764322917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,512,65536,0.12052693367004394
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,3072,0.01458453337351481
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,2560,0.004637866715590159
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,2560,0.014607999722162882
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,3584,0.012422399719556172
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,2048,0.0042357335488001505
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,2048,0.014467199643452963
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,3072,0.012123733758926392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,1536,0.0039381332695484165
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,2560,0.010840533177057902
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,2048,0.009928533434867859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,1536,0.014478933811187745
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,1024,0.0030048000315825146
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,1024,0.014542933305104574
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,768,0.003014400104681651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,1536,0.008821333448092144
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,1024,0.008322133123874665
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,768,0.01453439990679423
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,512,0.00258240004380544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,512,0.014390400052070618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,768,0.007971199850241344
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,256,0.002250666668017705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,512,0.007107200225194295
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,128,0.002169599880774816
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,256,0.013868799805641175
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,256,0.007197866837183635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,128,0.012921599547068277
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,64,0.002251733342806498
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,128,0.00625600020090739
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,256,32,0.0022762666145960493
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,64,0.013011200229326883
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,65536,0.009640533725420635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,256,32,0.013790933291117349
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,16384,0.005129600067933401
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,65536,0.01859626571337382
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,16384,0.01830613414446513
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,12288,0.004731733103593191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,12288,0.016720000902811685
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,16384,0.03453866640726726
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,12288,0.0276202658812205
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,256,65536,0.11975146929423015
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,10240,0.0054666668176651
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,10240,0.017778132359186807
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,8192,0.005355733136336008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,8192,0.017339734236399333
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,10240,0.02426346739133199
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,7168,0.005025066435337067
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,7168,0.01664959987004598
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,6144,0.004664533336957296
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,6144,0.0165503998597463
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,8192,0.020358399550120036
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,5120,0.004423466821511587
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,5120,0.016821332772572837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,7168,0.019142399231592812
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,4096,0.003961600114901861
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,6144,0.016990933815638223
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,4096,0.016707199811935424
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,3584,0.004740266501903534
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,3584,0.014618666966756186
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,5120,0.01539413332939148
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,4096,0.013312000036239623
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,3072,0.004310399790604909
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,3072,0.014458666245142618
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,2560,0.004605866471926371
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,2560,0.01446399986743927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,3584,0.012542933225631714
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,2048,0.0042250668009122215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,3072,0.011660800377527872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,2048,0.014493866761525472
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,65536,0.11978346506754559
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,1536,0.0033941333492596946
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,2560,0.010785067081451416
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,1024,0.0029653333127498626
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,1536,0.014594133694966635
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,2048,0.00957973301410675
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,768,0.0030026666820049284
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,1024,0.0134442667166392
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,1536,0.009118933478991191
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,768,0.01453013320763906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,512,0.0025802666942278544
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,1024,0.008366933465003968
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,256,0.0022026665508747103
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,768,0.007894399762153625
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,512,0.014424533645311991
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,128,0.0025237334271272023
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,512,0.007152000069618225
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,256,0.01404800017674764
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,64,0.0022122666239738466
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,128,0.014460800091425577
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,256,0.006680533289909363
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,128,32,0.002197333425283432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,nvfp4,1,128,128,0.006697600086530049
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,64,0.013330133756001792
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,65536,0.006663466493288676
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,128,32,0.012663466731707254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,16384,0.0050453335046768185
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,12288,0.005030400057633718
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,65536,0.019049600760142008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,16384,0.019026132424672444
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,10240,0.004643199841181437
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,12288,0.016794667641321818
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,8192,0.005542399982611338
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,7168,0.004646400113900503
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,10240,0.016530133287111917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,6144,0.004682666560014089
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,8192,0.01655359963575999
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,7168,0.017154133319854735
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,5120,0.004322133461634318
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,4096,0.004233600199222564
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,6144,0.01567893326282501
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,3584,0.005009066561857859
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,5120,0.016521599888801575
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,4096,0.016526933511098227
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,3072,0.004628266890843709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,3584,0.015381333231925965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,2560,0.00506986677646637
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,2048,0.0042303999265035
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,3072,0.014447999993960061
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,2560,0.015050666530927024
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,1536,0.003420799970626831
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,2048,0.014470400412877402
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,1024,0.0030239999294281008
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,768,0.002696533252795537
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,1536,0.014100266496340432
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,1024,0.01414293348789215
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,512,0.0026517334083716077
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,768,0.014468266566594442
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,256,0.0021568000316619872
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,128,0.0021845333278179167
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,512,0.01336853305498759
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,64,0.0021877333521842955
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,256,0.013338667154312134
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,128,0.01288106640179952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,64,32,0.0021920000513394673
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,65536,0.006657066444555919
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,64,0.01259519954522451
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,64,32,0.013185066978136697
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,16384,0.0050357331832249965
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,65536,0.01768640081087748
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,12288,0.005043200155099233
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,10240,0.005052799979845682
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,16384,0.017130666971206666
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,12288,0.016693333784739174
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,8192,0.004261333247025808
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,10240,0.016505600015322367
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,7168,0.004600533346335093
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,8192,0.01704960068066915
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,6144,0.004232533276081085
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,5120,0.004296533266703288
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,7168,0.016530133287111917
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,4096,0.004628266890843709
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,6144,0.016821332772572837
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,5120,0.016664533813794454
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,3584,0.004204800228277842
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,3072,0.003849600007136663
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,4096,0.016544000307718913
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,3584,0.014544000228246054
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,2560,0.004710400104522705
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,3072,0.014694399634997048
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,2048,0.004621866842110952
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,2560,0.014461867014567056
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,1536,0.003499733408292135
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,1024,0.003054933249950409
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,2048,0.014553599556287131
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,1536,0.014200533429781595
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,768,0.0026730666557947796
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,512,0.0025450666745503742
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,1024,0.01446399986743927
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,768,0.013065600395202636
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,256,0.0022154666483402254
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,128,0.0022005334496498106
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,512,0.014088533322016396
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,256,0.012583466370900473
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,64,0.0018602666755517324
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,float16,1,32,32,0.0020682666450738906
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,128,0.013222400347391763
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,64,0.012533332904179892
TRTLLM,1.1.0,NVIDIA GB200,gemm,torch_flow,fp8,1,32,32,0.014052266875902811
