framework,version,device,op_name,kernel_source,m,k,quant_dtype,latency
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,16384,fp8,0.2583807945251465
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,12288,fp8,0.19457173347473145
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,10240,fp8,0.1627914587656657
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,8192,fp8,0.13080639839172364
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,7168,fp8,0.11481813589731853
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,6144,fp8,0.09911039670308432
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,5120,fp8,0.08311786651611328
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,4096,fp8,0.06732587019602457
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,3584,fp8,0.05926506519317627
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,3072,fp8,0.051261866092681886
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,2560,fp8,0.04337386687596639
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,2048,fp8,0.033037867148717245
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,1536,fp8,0.018338133891423546
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,1024,fp8,0.01307093302408854
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,768,fp8,0.010071466366449993
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,512,fp8,0.007494399944941203
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,256,fp8,0.004704000055789947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,128,fp8,0.0037759999434153237
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,64,fp8,0.0032170665760835014
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,32,fp8,0.0030570665995279947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8192,65536,fp8,1.0362688064575196
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,16384,fp8,0.1305888017018636
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,12288,fp8,0.09903999964396158
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,10240,fp8,0.08286293347676596
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,8192,fp8,0.06702933311462403
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,7168,fp8,0.05913066864013672
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,6144,fp8,0.05119786659876505
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,5120,fp8,0.043397335211435954
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,4096,fp8,0.032330665985743204
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,3584,fp8,0.0235317329565684
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,3072,fp8,0.018557866414388023
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,2560,fp8,0.015435733397801719
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,2048,fp8,0.012831999858220419
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,1536,fp8,0.01004266639550527
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,1024,fp8,0.007472000022729237
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,768,fp8,0.006027733286221823
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,512,fp8,0.004701866706212362
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,256,fp8,0.0037461332976818085
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,128,fp8,0.0032000000278155005
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,64,fp8,0.003053866575161616
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,32,fp8,0.0028437333802382152
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4096,65536,fp8,0.5159029324849447
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,16384,fp8,0.06724906762440999
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,12288,fp8,0.051292800903320314
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,10240,fp8,0.04342079957326253
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,8192,fp8,0.03316906690597534
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,7168,fp8,0.022680532932281495
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,6144,fp8,0.018415999412536622
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,5120,fp8,0.015453867117563882
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,4096,fp8,0.012850133577982583
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,3584,fp8,0.011411199967066448
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,3072,fp8,0.010166399677594503
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,2560,fp8,0.008840533097585042
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,2048,fp8,0.0073728000124295544
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,1536,fp8,0.006107733150323232
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,1024,fp8,0.004704000055789947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,768,fp8,0.004127999891837438
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,512,fp8,0.0037600000699361167
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,256,fp8,0.0032042667269706728
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,128,fp8,0.003036800026893616
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,64,fp8,0.0028266665836175283
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,32,fp8,0.0027093333502610523
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2048,65536,fp8,0.25882986386617024
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,16384,fp8,0.03208960096041362
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,12288,fp8,0.01827413241068522
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,10240,fp8,0.015503999590873719
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,8192,fp8,0.01272640029589335
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,7168,fp8,0.01151039997736613
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,6144,fp8,0.010179199775060018
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,5120,fp8,0.008705066641171773
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,4096,fp8,0.007468800246715546
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,3584,fp8,0.006715733309586842
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,3072,fp8,0.006045866509278615
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,2560,fp8,0.005426133175690969
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,2048,fp8,0.004705066482226053
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,1536,fp8,0.004227200150489807
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,1024,fp8,0.003755733370780945
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,768,fp8,0.0034229333202044168
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,512,fp8,0.003291733314593633
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,256,fp8,0.0029557332396507262
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,128,fp8,0.002833066632350286
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,64,fp8,0.0027104000250498454
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,32,fp8,0.002810666710138321
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1024,65536,fp8,0.13069120248158772
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,16384,fp8,0.018293333053588868
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,12288,fp8,0.01411733329296112
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,10240,fp8,0.012065066893895467
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,8192,fp8,0.010161067048708599
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,7168,fp8,0.00918933351834615
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,6144,fp8,0.008087466657161712
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,5120,fp8,0.007159466544787089
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,4096,fp8,0.006040533383687338
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,3584,fp8,0.005460266768932342
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,3072,fp8,0.0050901333491007485
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,2560,fp8,0.004589866598447164
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,2048,fp8,0.004223999877770742
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,1536,fp8,0.0038794666528701783
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,1024,fp8,0.0035690667728583017
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,768,fp8,0.003323733309904734
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,512,fp8,0.0031989333530267083
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,256,fp8,0.0029386666913827257
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,128,fp8,0.0027882667879263563
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,64,fp8,0.0028021333118279776
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,32,fp8,0.002703999976317088
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,768,65536,fp8,0.0990015983581543
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,16384,fp8,0.01285866697629293
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,12288,fp8,0.01009493370850881
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,10240,fp8,0.008835200468699138
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,8192,fp8,0.007500799993673961
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,7168,fp8,0.006718933085600535
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,6144,fp8,0.0061471998691558834
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,5120,fp8,0.005381333331267038
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,4096,fp8,0.004822400212287903
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,3584,fp8,0.004385066529115042
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,3072,fp8,0.004240000247955322
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,2560,fp8,0.0039018665750821433
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,2048,fp8,0.003688533355792364
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,1536,fp8,0.003516799956560135
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,1024,fp8,0.0032085334261258446
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,768,fp8,0.0032085334261258446
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,512,fp8,0.003065599997838338
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,256,fp8,0.0027615999182065325
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,128,fp8,0.0028149334092934927
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,64,fp8,0.002717866748571396
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,32,fp8,0.002713600049416224
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,512,65536,fp8,0.06738346417744955
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,16384,fp8,0.010070400436719258
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,12288,fp8,0.008155733346939087
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,10240,fp8,0.007149866720040639
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,8192,fp8,0.006010666489601135
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,7168,fp8,0.005545599758625031
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,6144,fp8,0.005003733436266581
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,5120,fp8,0.004586666822433472
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,4096,fp8,0.004203733305136363
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,3584,fp8,0.004008533308903376
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,3072,fp8,0.003942399968703588
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,2560,fp8,0.0037493333220481873
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,2048,fp8,0.0034133332471052804
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,1536,fp8,0.0033717334270477297
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,1024,fp8,0.0030826665461063385
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,768,fp8,0.002998399982849757
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,512,fp8,0.0028607999285062153
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,256,fp8,0.0027530667682488757
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,128,fp8,0.002799999962250392
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,64,fp8,0.0027232001225153605
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,32,fp8,0.0027797333896160126
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,384,65536,fp8,0.05124693314234415
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,16384,fp8,0.007479466497898102
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,12288,fp8,0.006110933423042297
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,10240,fp8,0.005336533486843109
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,8192,fp8,0.004783999919891357
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,7168,fp8,0.004365866879622141
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,6144,fp8,0.004124800115823746
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,5120,fp8,0.004004266609748204
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,4096,fp8,0.0036650667587916053
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,3584,fp8,0.0036799999574820197
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,3072,fp8,0.0035146666069825493
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,2560,fp8,0.0035818666219711304
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,2048,fp8,0.0033013333876927694
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,1536,fp8,0.003101866692304611
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,1024,fp8,0.003050666550795237
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,768,fp8,0.002841600030660629
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,512,fp8,0.002863999952872594
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,256,fp8,0.002726399898529053
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,128,fp8,0.00271573339899381
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,64,fp8,0.0027818667391935987
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,32,fp8,0.002696533252795537
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,256,65536,fp8,0.03296639919281006
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,16384,fp8,0.0061258668700853985
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,12288,fp8,0.005021866659323374
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,10240,fp8,0.004683733483155568
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,8192,fp8,0.004153600086768469
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,7168,fp8,0.004051200052102407
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,6144,fp8,0.003942399968703588
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,5120,fp8,0.003659733384847641
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,4096,fp8,0.0035349334279696143
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,3584,fp8,0.0034048000971476236
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,3072,fp8,0.003398400048414866
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,2560,fp8,0.0031957333286603295
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,2048,fp8,0.0031946666538715364
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,1536,fp8,0.0030879999200503034
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,1024,fp8,0.0028138667345046996
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,768,fp8,0.0028405333558718365
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,512,fp8,0.0027456000447273255
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,256,fp8,0.0027242665489514667
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,128,fp8,0.002807466685771942
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,64,fp8,0.002681600054105123
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,32,fp8,0.0027882667879263563
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,192,65536,fp8,0.018346667289733887
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,16384,fp8,0.005350400010744731
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,12288,fp8,0.004679466784000397
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,10240,fp8,0.004259199897448222
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,8192,fp8,0.003924266745646795
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,7168,fp8,0.003942399968703588
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,6144,fp8,0.0036618667344252265
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,5120,fp8,0.003576533248027166
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,4096,fp8,0.0034517332911491393
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,3584,fp8,0.0032640000184377036
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,3072,fp8,0.0032821332414944967
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,2560,fp8,0.003155199935038885
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,2048,fp8,0.003047466774781545
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,1536,fp8,0.0029898665845394133
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,1024,fp8,0.0027797333896160126
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,768,fp8,0.0028618666032950084
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,512,fp8,0.002825599908828735
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,256,fp8,0.002735999971628189
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,128,fp8,0.00279573326309522
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,64,fp8,0.0027104000250498454
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,32,fp8,0.0027872001131375628
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,160,65536,fp8,0.015398400028546652
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,16384,fp8,0.004796800017356872
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,12288,fp8,0.0041685332854588825
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,10240,fp8,0.003922133396069208
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,8192,fp8,0.003764266769091288
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,7168,fp8,0.003572266548871994
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,6144,fp8,0.0035402665535608927
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,5120,fp8,0.00345920001467069
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,4096,fp8,0.0031957333286603295
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,3584,fp8,0.003269333392381668
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,3072,fp8,0.003099733342727025
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,2560,fp8,0.003053866575161616
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,2048,fp8,0.0030570665995279947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,1536,fp8,0.0028319999575614927
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,1024,fp8,0.002890666574239731
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,768,fp8,0.002776533365249634
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,512,fp8,0.0028192001084486645
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,256,fp8,0.0026922665536403658
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,128,fp8,0.002784000088771184
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,64,fp8,0.002773333340883255
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,32,fp8,0.002681600054105123
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,128,65536,fp8,0.012824533383051553
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,16384,fp8,0.004136533290147781
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,12288,fp8,0.0038549333810806276
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,10240,fp8,0.003739733248949051
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,8192,fp8,0.0034282666941483817
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,7168,fp8,0.0037109332780043284
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,6144,fp8,0.003368533402681351
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,5120,fp8,0.003186133255561193
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,4096,fp8,0.0031871999303499854
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,3584,fp8,0.0030591999491055804
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,3072,fp8,0.003089066594839096
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,2560,fp8,0.0029045333464940387
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,2048,fp8,0.0029130667448043824
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,1536,fp8,0.0027402666707833606
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,1024,fp8,0.0027445333699385325
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,768,fp8,0.002811733384927114
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,512,fp8,0.002717866748571396
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,256,fp8,0.002791466563940048
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,128,fp8,0.002776533365249634
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,64,fp8,0.002679466704527537
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,32,fp8,0.0027722666660944624
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,96,65536,fp8,0.010070400436719258
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,16384,fp8,0.0038986665507157645
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,12288,fp8,0.003738666574160258
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,10240,fp8,0.003457066665093104
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,8192,fp8,0.0034506666163603462
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,7168,fp8,0.0032416000962257386
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,6144,fp8,0.003271466741959254
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,5120,fp8,0.0031093334158261614
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,4096,fp8,0.0031285333136717476
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,3584,fp8,0.0030602666238943735
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,3072,fp8,0.002881066749493281
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,2560,fp8,0.002883200099070867
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,2048,fp8,0.0027637332677841187
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,1536,fp8,0.002739199995994568
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,1024,fp8,0.0028138667345046996
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,768,fp8,0.0026933332284291584
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,512,fp8,0.0027978666126728057
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,256,fp8,0.00277866671482722
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,128,fp8,0.0026933332284291584
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,64,fp8,0.0027722666660944624
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,32,fp8,0.002679466704527537
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,80,65536,fp8,0.00883840024471283
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,16384,fp8,0.003655466685692469
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,12288,fp8,0.0038560000558694207
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,10240,fp8,0.003345066557327906
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,8192,fp8,0.0031968000034491217
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,7168,fp8,0.0032416000962257386
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,6144,fp8,0.003101866692304611
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,5120,fp8,0.003123199939727783
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,4096,fp8,0.0030570665995279947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,3584,fp8,0.0028362666567166646
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,3072,fp8,0.0028970666229724885
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,2560,fp8,0.002765866617361705
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,2048,fp8,0.0027456000447273255
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,1536,fp8,0.002833066632350286
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,1024,fp8,0.0027072000006834666
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,768,fp8,0.002820266783237457
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,512,fp8,0.0027306665976842242
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,256,fp8,0.0028021333118279776
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,128,fp8,0.0026954665780067446
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,64,fp8,0.003094399968783061
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,32,fp8,0.0028010666370391845
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,64,65536,fp8,0.007379200061162312
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,16384,fp8,0.0035232000052928926
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,12288,fp8,0.003311999887228012
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,10240,fp8,0.003193599979082743
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,8192,fp8,0.0032085334261258446
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,7168,fp8,0.003053866575161616
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,6144,fp8,0.00308693324526151
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,5120,fp8,0.0029824001093705496
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,4096,fp8,0.002808533360560735
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,3584,fp8,0.002868266652027766
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,3072,fp8,0.0027594665686289472
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,2560,fp8,0.0028490667541821797
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,2048,fp8,0.0027669332921504974
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,1536,fp8,0.002833066632350286
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,1024,fp8,0.002733866622050603
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,768,fp8,0.0027082666754722597
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,512,fp8,0.00279573326309522
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,256,fp8,0.0026954665780067446
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,128,fp8,0.002794666588306427
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,64,fp8,0.002799999962250392
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,32,fp8,0.002699733277161916
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,48,65536,fp8,0.0061365331212679545
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,16384,fp8,0.003218133250872294
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,12288,fp8,0.0031008000175158186
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,10240,fp8,0.003139200061559677
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,8192,fp8,0.002993066608905792
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,7168,fp8,0.0029493334392706556
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,6144,fp8,0.002899199972550074
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,5120,fp8,0.0027647999425729113
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,4096,fp8,0.0028512001037597655
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,3584,fp8,0.002757333219051361
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,3072,fp8,0.002754133443037669
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,2560,fp8,0.0028160000840822858
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,2048,fp8,0.002712533374627431
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,1536,fp8,0.0028053333361943563
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,1024,fp8,0.0027935999135176343
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,768,fp8,0.0027232001225153605
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,512,fp8,0.0028042666614055633
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,256,fp8,0.0027029333015282948
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,128,fp8,0.0026911998788515727
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,64,fp8,0.003019733230272929
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,32,fp8,0.002696533252795537
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,32,65536,fp8,0.0048096001148223875
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,16384,fp8,0.0029919999341169994
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,12288,fp8,0.0028768000503381092
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,10240,fp8,0.0027829334139823914
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,8192,fp8,0.0028490667541821797
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,7168,fp8,0.002833066632350286
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,6144,fp8,0.00277866671482722
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,5120,fp8,0.0028213332096735638
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,4096,fp8,0.0027456000447273255
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,3584,fp8,0.0027146667242050173
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,3072,fp8,0.002845866729815801
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,2560,fp8,0.0027295999228954316
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,2048,fp8,0.0028138667345046996
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,1536,fp8,0.0028031999866167706
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,1024,fp8,0.0027104000250498454
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,768,fp8,0.0027893332143624624
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,512,fp8,0.0026687999566396077
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,256,fp8,0.0027935999135176343
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,128,fp8,0.002681600054105123
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,64,fp8,0.0027744000156720476
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,32,fp8,0.002674133330583572
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,16,65536,fp8,0.003651199986537298
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,16384,fp8,0.002850133428970973
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,12288,fp8,0.0027424000203609467
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,10240,fp8,0.0028149334092934927
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,8192,fp8,0.0028234665592511495
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,7168,fp8,0.0027253332237402597
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,6144,fp8,0.002820266783237457
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,5120,fp8,0.002705066651105881
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,4096,fp8,0.0027072000006834666
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,3584,fp8,0.0027797333896160126
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,3072,fp8,0.002700799951950709
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,2560,fp8,0.0027829334139823914
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,2048,fp8,0.002690133452415466
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,1536,fp8,0.0027882667879263563
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,1024,fp8,0.0026943999032179515
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,768,fp8,0.0027637332677841187
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,512,fp8,0.0027744000156720476
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,256,fp8,0.002669866631428401
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,128,fp8,0.0027722666660944624
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,64,fp8,0.002664533257484436
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,32,fp8,0.0027061333258946735
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,8,65536,fp8,0.0033141332368055976
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,16384,fp8,0.0027242665489514667
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,12288,fp8,0.002810666710138321
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,10240,fp8,0.002811733384927114
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,8192,fp8,0.002700799951950709
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,7168,fp8,0.002796799937884013
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,6144,fp8,0.0027445333699385325
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,5120,fp8,0.0026880001028378804
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,4096,fp8,0.0027850667635599775
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,3584,fp8,0.002701866626739502
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,3072,fp8,0.002807466685771942
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,2560,fp8,0.002784000088771184
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,2048,fp8,0.002705066651105881
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,1536,fp8,0.0027776000400384264
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,1024,fp8,0.002712533374627431
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,768,fp8,0.0026933332284291584
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,512,fp8,0.002792533238728841
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,256,fp8,0.0026880001028378804
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,128,fp8,0.002773333340883255
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,64,fp8,0.0027829334139823914
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,32,fp8,0.0027114666998386385
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,4,65536,fp8,0.0030634666482607523
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,16384,fp8,0.0027072000006834666
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,12288,fp8,0.0028053333361943563
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,10240,fp8,0.0027146667242050173
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,8192,fp8,0.0028031999866167706
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,7168,fp8,0.0026986666023731233
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,6144,fp8,0.0026890667776266735
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,5120,fp8,0.0027850667635599775
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,4096,fp8,0.002696533252795537
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,3584,fp8,0.0027818667391935987
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,3072,fp8,0.0027882667879263563
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,2560,fp8,0.002705066651105881
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,2048,fp8,0.0027850667635599775
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,1536,fp8,0.0026858667532602947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,1024,fp8,0.0026986666023731233
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,768,fp8,0.0028021333118279776
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,512,fp8,0.0026986666023731233
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,256,fp8,0.002776533365249634
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,128,fp8,0.0026858667532602947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,64,fp8,0.0027701333165168762
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,32,fp8,0.002674133330583572
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,2,65536,fp8,0.002844800055027008
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,16384,fp8,0.0027903998891512555
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,12288,fp8,0.0026837334036827086
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,10240,fp8,0.0027829334139823914
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,8192,fp8,0.0026762666801611584
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,7168,fp8,0.0026848000784715016
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,6144,fp8,0.0027669332921504974
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,5120,fp8,0.0026762666801611584
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,4096,fp8,0.002791466563940048
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,3584,fp8,0.0027776000400384264
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,3072,fp8,0.0026837334036827086
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,2560,fp8,0.00276053324341774
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,2048,fp8,0.0026933332284291584
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,1536,fp8,0.00277866671482722
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,1024,fp8,0.0026858667532602947
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,768,fp8,0.0027594665686289472
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,512,fp8,0.002669866631428401
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,256,fp8,0.002678400029738744
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,128,fp8,0.002775466690460841
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,64,fp8,0.0026880001028378804
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,32,fp8,0.0028010666370391845
TRTLLM,1.0.0rc3,NVIDIA H200,scale_matrix,torch_ops,1,65536,fp8,0.002820266783237457
