framework,version,device,op_name,kernel_source,m,k,quant_dtype,latency
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,16384,fp8,0.2724671999613444
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,12288,fp8,0.2060544013977051
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,10240,fp8,0.17242560386657715
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,8192,fp8,0.13810985883076984
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,7168,fp8,0.12095359961191814
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,6144,fp8,0.10445546309153239
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,5120,fp8,0.08784533341725667
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,4096,fp8,0.056184534231821695
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,3584,fp8,0.035499731699625656
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,3072,fp8,0.029190399249394733
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,2560,fp8,0.02480640014012655
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,2048,fp8,0.020678400993347168
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,1536,fp8,0.01611733337243398
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,1024,fp8,0.011987200379371643
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,768,fp8,0.00965333382288615
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,512,fp8,0.007100800176461537
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,256,fp8,0.005062399804592133
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,128,fp8,0.003939199944337209
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,64,fp8,0.0034474665919939675
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,32,fp8,0.0029781334102153777
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8192,65536,fp8,1.115936024983724
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,16384,fp8,0.14578240712483723
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,12288,fp8,0.1109887997309367
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,10240,fp8,0.09420479933420817
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,8192,fp8,0.06240640083948771
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,7168,fp8,0.044047999382019046
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,6144,fp8,0.03573973178863525
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,5120,fp8,0.03126613299051921
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,4096,fp8,0.02600319981575012
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,3584,fp8,0.022765866915384927
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,3072,fp8,0.02013333241144816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,2560,fp8,0.0173909326394399
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,2048,fp8,0.01446293294429779
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,1536,fp8,0.011713066697120666
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,1024,fp8,0.008724266290664673
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,768,fp8,0.0070720002055168155
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,512,fp8,0.005857066810131073
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,256,fp8,0.004222933451334635
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,128,fp8,0.0038058665891488397
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,64,fp8,0.0033919999996821085
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,32,fp8,0.002974933385848999
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4096,65536,fp8,1.0995455423990885
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,16384,fp8,0.1441696008046468
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,12288,fp8,0.09808320204416911
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,10240,fp8,0.07097919782002768
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,8192,fp8,0.05192213455835978
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,7168,fp8,0.043961600462595625
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,6144,fp8,0.03875946601231893
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,5120,fp8,0.029550933837890626
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,4096,fp8,0.024235733350118003
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,3584,fp8,0.021817600727081297
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,3072,fp8,0.01774079998334249
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,2560,fp8,0.015280000368754067
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,2048,fp8,0.013218133648236593
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,1536,fp8,0.011106133460998535
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,1024,fp8,0.008595200379689534
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,768,fp8,0.007097599903742473
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,512,fp8,0.005422933399677277
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,256,fp8,0.004649599889914194
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,128,fp8,0.00420053352912267
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,64,fp8,0.0034058667719364167
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,32,fp8,0.0029813334345817565
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2048,65536,fp8,0.46892480850219725
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,16384,fp8,0.06322346528371176
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,12288,fp8,0.048987734317779544
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,10240,fp8,0.03316906690597534
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,8192,fp8,0.02765760024388631
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,7168,fp8,0.02435520092646281
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,6144,fp8,0.018518400192260743
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,5120,fp8,0.015260799725850423
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,4096,fp8,0.01323306659857432
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,3584,fp8,0.013142399986584983
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,3072,fp8,0.009130666653315227
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,2560,fp8,0.009133866429328919
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,2048,fp8,0.007492266595363617
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,1536,fp8,0.006949333349863689
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,1024,fp8,0.0054954667886098225
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,768,fp8,0.005433600147565206
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,512,fp8,0.0038015998899936674
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,256,fp8,0.003519999980926514
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,128,fp8,0.0029877332349618276
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,64,fp8,0.0025568000972270967
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,32,fp8,0.0025813333690166474
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1024,65536,fp8,0.14360960324605304
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,16384,fp8,0.018422400951385497
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,12288,fp8,0.01446399986743927
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,10240,fp8,0.01322879989941915
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,8192,fp8,0.010678399602572124
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,7168,fp8,0.01564800043900808
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,6144,fp8,0.014061866203943887
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,5120,fp8,0.012684800227483115
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,4096,fp8,0.010373333096504211
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,3584,fp8,0.00995199978351593
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,3072,fp8,0.009514666597048442
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,2560,fp8,0.006413866579532623
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,2048,fp8,0.005857066810131073
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,1536,fp8,0.005017599960168203
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,1024,fp8,0.0042133331298828125
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,768,fp8,0.0037930667400360107
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,512,fp8,0.0034122665723164878
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,256,fp8,0.0025631998976071674
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,128,fp8,0.002583466718594233
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,64,fp8,0.002573866645495097
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,32,fp8,0.0021546666820844015
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,768,65536,fp8,0.10443840026855469
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,16384,fp8,0.011595732967058818
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,12288,fp8,0.009516800443331402
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,10240,fp8,0.008725333213806152
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,8192,fp8,0.007166933516661327
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,7168,fp8,0.0070933332045873005
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,6144,fp8,0.006257066627343495
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,5120,fp8,0.005834666887919108
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,4096,fp8,0.005036800106366476
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,3584,fp8,0.005437866846720377
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,3072,fp8,0.005029333134492239
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,2560,fp8,0.00418453315893809
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,2048,fp8,0.0038730666041374207
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,1536,fp8,0.0038143999874591826
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,1024,fp8,0.0037237333754698435
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,768,fp8,0.0033941333492596946
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,512,fp8,0.002921599894762039
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,256,fp8,0.0029343999922275543
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,128,fp8,0.0025685332715511323
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,64,fp8,0.002160000056028366
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,32,fp8,0.002598399917284648
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,512,65536,fp8,0.05788799921671549
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,16384,fp8,0.009545600414276123
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,12288,fp8,0.008317866424719492
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,10240,fp8,0.007484800120194752
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,8192,fp8,0.006337066491444905
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,7168,fp8,0.006230400005976359
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,6144,fp8,0.005449600021044413
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,5120,fp8,0.005092266698678335
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,4096,fp8,0.004636799792448679
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,3584,fp8,0.004194133480389913
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,3072,fp8,0.004205866654713949
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,2560,fp8,0.003734400123357773
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,2048,fp8,0.00421013335386912
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,1536,fp8,0.0037546666959921518
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,1024,fp8,0.0030026666820049284
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,768,fp8,0.003032533327738444
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,512,fp8,0.0025770666698614756
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,256,fp8,0.002569599946339925
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,128,fp8,0.002570666621128718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,64,fp8,0.002552533398071925
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,32,fp8,0.002165333429972331
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,384,65536,fp8,0.029209599892298384
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,16384,fp8,0.007482666770617168
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,12288,fp8,0.006258133550484974
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,10240,fp8,0.006251733501752217
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,8192,fp8,0.005027199784914652
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,7168,fp8,0.005022933085759481
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,6144,fp8,0.005038933455944061
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,5120,fp8,0.00421973317861557
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,4096,fp8,0.003807999938726425
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,3584,fp8,0.0038005332152048744
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,3072,fp8,0.003804799914360046
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,2560,fp8,0.003788800040880839
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,2048,fp8,0.0034154665966828666
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,1536,fp8,0.0029909332593282064
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,1024,fp8,0.0029696000119050344
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,768,fp8,0.0025781333446502686
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,512,fp8,0.0029472000896930695
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,256,fp8,0.002570666621128718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,128,fp8,0.0021664001047611236
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,64,fp8,0.0025546667476495106
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,32,fp8,0.002605866640806198
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,256,65536,fp8,0.020631466309229532
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,16384,fp8,0.006250666578610738
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,12288,fp8,0.005419733126958212
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,10240,fp8,0.005025066435337067
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,8192,fp8,0.005030400057633718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,7168,fp8,0.004222933451334635
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,6144,fp8,0.004206933577855428
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,5120,fp8,0.003815466662247976
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,4096,fp8,0.0037823999921480812
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,3584,fp8,0.0034613333642482757
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,3072,fp8,0.0038005332152048744
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,2560,fp8,0.003389866650104523
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,2048,fp8,0.002998399982849757
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,1536,fp8,0.0030432000756263735
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,1024,fp8,0.0025802666942278544
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,768,fp8,0.0025727999707063037
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,512,fp8,0.0025589334468046824
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,256,fp8,0.0021664001047611236
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,128,fp8,0.0025653332471847535
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,64,fp8,0.0025589334468046824
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,32,fp8,0.0021877333521842955
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,192,65536,fp8,0.01650879979133606
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,16384,fp8,0.00586346685886383
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,12288,fp8,0.005021866659323374
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,10240,fp8,0.0050005331635475155
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,8192,fp8,0.004565333326657614
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,7168,fp8,0.003806933263937632
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,6144,fp8,0.003787733366092046
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,5120,fp8,0.0038751999537150065
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,4096,fp8,0.0033887999753157297
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,3584,fp8,0.00378560001651446
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,3072,fp8,0.003403733422358831
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,2560,fp8,0.0029706666866938275
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,2048,fp8,0.003389866650104523
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,1536,fp8,0.002921599894762039
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,1024,fp8,0.0026047999660174055
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,768,fp8,0.0025610665480295816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,512,fp8,0.0026079999903837843
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,256,fp8,0.0025642665723959604
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,128,fp8,0.002583466718594233
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,64,fp8,0.002161066730817159
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,32,fp8,0.002164266755183538
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,160,65536,fp8,0.014107732971509298
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,16384,fp8,0.005047466854254405
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,12288,fp8,0.004600533346335093
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,10240,fp8,0.0045621335506439206
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,8192,fp8,0.0037994667887687682
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,7168,fp8,0.003755733370780945
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,6144,fp8,0.004080000023047129
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,5120,fp8,0.0034005333979924522
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,4096,fp8,0.0034143999218940735
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,3584,fp8,0.0033962666988372804
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,3072,fp8,0.003005866706371307
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,2560,fp8,0.0033333333830038703
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,2048,fp8,0.0029418667157491045
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,1536,fp8,0.00257493332028389
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,1024,fp8,0.0029898665845394133
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,768,fp8,0.0025653332471847535
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,512,fp8,0.0025546667476495106
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,256,fp8,0.0025781333446502686
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,128,fp8,0.002163200080394745
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,64,fp8,0.002518400053183238
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,32,fp8,0.002162133405605952
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,128,65536,fp8,0.011590400338172912
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,16384,fp8,0.005036800106366476
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,12288,fp8,0.003803733239571253
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,10240,fp8,0.003822933385769526
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,8192,fp8,0.003930666546026865
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,7168,fp8,0.0034048000971476236
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,6144,fp8,0.003790933390458425
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,5120,fp8,0.0033418667813142145
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,4096,fp8,0.002995199958483378
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,3584,fp8,0.003389866650104523
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,3072,fp8,0.003045333425203959
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,2560,fp8,0.002980266759792964
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,2048,fp8,0.0025631998976071674
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,1536,fp8,0.002926933268706004
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,1024,fp8,0.002571733295917511
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,768,fp8,0.0025568000972270967
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,512,fp8,0.002586666742960612
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,256,fp8,0.0021536000072956084
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,128,fp8,0.002549333373705546
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,64,fp8,0.002182399978240331
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,32,fp8,0.0021407999098300933
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,96,65536,fp8,0.009810133775075277
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,16384,fp8,0.004204800228277842
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,12288,fp8,0.003803733239571253
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,10240,fp8,0.003924266745646795
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,8192,fp8,0.0034048000971476236
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,7168,fp8,0.003619199991226196
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,6144,fp8,0.003386666625738144
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,5120,fp8,0.0029813334345817565
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,4096,fp8,0.003336533407370249
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,3584,fp8,0.002985599885384242
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,3072,fp8,0.0029898665845394133
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,2560,fp8,0.0029898665845394133
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,2048,fp8,0.002570666621128718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,1536,fp8,0.002566399921973546
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,1024,fp8,0.0025066666305065153
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,768,fp8,0.0023530667026837667
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,512,fp8,0.0025802666942278544
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,256,fp8,0.0021578667064507803
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,128,fp8,0.0025888000925381976
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,64,fp8,0.0025600001215934755
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,32,fp8,0.002164266755183538
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,80,65536,fp8,0.008345599969228108
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,16384,fp8,0.0038090666135152185
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,12288,fp8,0.004163199911514918
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,10240,fp8,0.003401600072781245
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,8192,fp8,0.0033791999022165934
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,7168,fp8,0.0029813334345817565
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,6144,fp8,0.0029898665845394133
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,5120,fp8,0.003387733300526937
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,4096,fp8,0.002976000060637792
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,3584,fp8,0.0029834667841593427
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,3072,fp8,0.002962133288383484
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,2560,fp8,0.0025589334468046824
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,2048,fp8,0.0029781334102153777
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,1536,fp8,0.002499199906984965
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,1024,fp8,0.0021738665799299877
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,768,fp8,0.0025568000972270967
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,512,fp8,0.0022111999491850535
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,256,fp8,0.0025759999950726825
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,128,fp8,0.002586666742960612
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,64,fp8,0.002148266633351644
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,32,fp8,0.002566399921973546
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,64,65536,fp8,0.007218133409818013
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,16384,fp8,0.0038058665891488397
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,12288,fp8,0.00378560001651446
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,10240,fp8,0.003386666625738144
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,8192,fp8,0.0033813332517941795
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,7168,fp8,0.0033258666594823206
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,6144,fp8,0.00297173336148262
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,5120,fp8,0.002995199958483378
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,4096,fp8,0.0025813333690166474
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,3584,fp8,0.002600533266862233
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,3072,fp8,0.0029279999434947968
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,2560,fp8,0.0025642665723959604
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,2048,fp8,0.0025162667036056517
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,1536,fp8,0.0025087999800841015
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,1024,fp8,0.002551466723283132
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,768,fp8,0.0021546666820844015
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,512,fp8,0.002571733295917511
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,256,fp8,0.002569599946339925
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,128,fp8,0.002162133405605952
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,64,fp8,0.002583466718594233
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,32,fp8,0.002164266755183538
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,48,65536,fp8,0.006215466558933258
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,16384,fp8,0.0034624000390370687
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,12288,fp8,0.0029887999097506206
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,10240,fp8,0.0033717334270477297
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,8192,fp8,0.0030261332790056865
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,7168,fp8,0.0029813334345817565
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,6144,fp8,0.0025642665723959604
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,5120,fp8,0.0029706666866938275
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,4096,fp8,0.0029279999434947968
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,3584,fp8,0.0025610665480295816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,3072,fp8,0.0025759999950726825
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,2560,fp8,0.0025589334468046824
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,2048,fp8,0.002216533323129018
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,1536,fp8,0.002550400048494339
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,1024,fp8,0.002152533332506816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,768,fp8,0.002585600068171819
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,512,fp8,0.0025770666698614756
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,256,fp8,0.002170666555563609
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,128,fp8,0.0025546667476495106
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,64,fp8,0.002162133405605952
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,32,fp8,0.002169599880774816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,32,65536,fp8,0.005046399931112925
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,16384,fp8,0.0029824001093705496
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,12288,fp8,0.0025568000972270967
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,10240,fp8,0.002573866645495097
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,8192,fp8,0.002977066735426585
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,7168,fp8,0.0025120000044504803
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,6144,fp8,0.0025578667720158894
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,5120,fp8,0.0025685332715511323
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,4096,fp8,0.002243199944496155
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,3584,fp8,0.0025802666942278544
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,3072,fp8,0.0021503999829292296
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,2560,fp8,0.0021461332837740582
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,2048,fp8,0.002553600072860718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,1536,fp8,0.002165333429972331
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,1024,fp8,0.002566399921973546
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,768,fp8,0.002553600072860718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,512,fp8,0.0021546666820844015
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,256,fp8,0.002553600072860718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,128,fp8,0.002178133279085159
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,64,fp8,0.002162133405605952
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,32,fp8,0.002585600068171819
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,16,65536,fp8,0.0038090666135152185
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,16384,fp8,0.002921599894762039
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,12288,fp8,0.002584533393383026
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,10240,fp8,0.0025216000775496166
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,8192,fp8,0.0025685332715511323
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,7168,fp8,0.002217599997917811
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,6144,fp8,0.002419200042883555
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,5120,fp8,0.0025792000194390613
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,4096,fp8,0.002169599880774816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,3584,fp8,0.002569599946339925
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,3072,fp8,0.002585600068171819
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,2560,fp8,0.0021578667064507803
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,2048,fp8,0.0021536000072956084
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,1536,fp8,0.0025813333690166474
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,1024,fp8,0.0025621332228183747
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,768,fp8,0.0021930667261282604
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,512,fp8,0.002587733417749405
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,256,fp8,0.002161066730817159
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,128,fp8,0.002181333303451538
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,64,fp8,0.002566399921973546
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,32,fp8,0.0021674667795499166
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,8,65536,fp8,0.0035285333792368567
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,16384,fp8,0.002571733295917511
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,12288,fp8,0.0021461332837740582
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,10240,fp8,0.0025770666698614756
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,8192,fp8,0.002162133405605952
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,7168,fp8,0.002152533332506816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,6144,fp8,0.0025077333052953084
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,5120,fp8,0.0021664001047611236
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,4096,fp8,0.002569599946339925
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,3584,fp8,0.0025610665480295816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,3072,fp8,0.0021568000316619872
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,2560,fp8,0.0025781333446502686
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,2048,fp8,0.0021557333568731946
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,1536,fp8,0.002164266755183538
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,1024,fp8,0.0025781333446502686
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,768,fp8,0.0021664001047611236
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,512,fp8,0.00257493332028389
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,256,fp8,0.002567466596762339
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,128,fp8,0.002170666555563609
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,64,fp8,0.0025248001019159954
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,32,fp8,0.002169599880774816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,4,65536,fp8,0.0029418667157491045
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,16384,fp8,0.002571733295917511
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,12288,fp8,0.002533333251873652
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,10240,fp8,0.002152533332506816
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,8192,fp8,0.002147199958562851
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,7168,fp8,0.0025642665723959604
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,6144,fp8,0.002164266755183538
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,5120,fp8,0.002549333373705546
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,4096,fp8,0.002566399921973546
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,3584,fp8,0.002165333429972331
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,3072,fp8,0.002593066543340683
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,2560,fp8,0.0021568000316619872
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,2048,fp8,0.002182399978240331
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,1536,fp8,0.0025759999950726825
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,1024,fp8,0.002149333308140437
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,768,fp8,0.002553600072860718
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,512,fp8,0.0025909334421157838
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,256,fp8,0.0021674667795499166
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,128,fp8,0.0025770666698614756
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,64,fp8,0.0021738665799299877
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,32,fp8,0.0021738665799299877
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,2,65536,fp8,0.002935466667016347
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,16384,fp8,0.0021738665799299877
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,12288,fp8,0.002583466718594233
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,10240,fp8,0.0025770666698614756
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,8192,fp8,0.0021514666577180227
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,7168,fp8,0.0025685332715511323
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,6144,fp8,0.0021685334543387097
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,5120,fp8,0.0021856000026067098
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,4096,fp8,0.002584533393383026
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,3584,fp8,0.002165333429972331
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,3072,fp8,0.00257493332028389
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,2560,fp8,0.0025685332715511323
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,2048,fp8,0.0021727999051411946
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,1536,fp8,0.002195200075705846
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,1024,fp8,0.00257493332028389
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,768,fp8,0.002593066543340683
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,512,fp8,0.0021546666820844015
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,256,fp8,0.00258240004380544
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,128,fp8,0.0021546666820844015
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,64,fp8,0.0021749332547187803
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,32,fp8,0.00258240004380544
TRTLLM,1.2.0rc5,NVIDIA B200,scale_matrix,torch_ops,1,65536,fp8,0.0021749332547187803
