framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,128,0.0072223998606204985
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,32,0.004860800132155419
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,16,0.004899200052022934
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,8,0.004687999933958053
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,4,0.004854400083422661
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,2,0.0047391999512910845
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,1,0.004902400076389313
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,128,0.007248000055551529
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,64,0.006364800035953522
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,32,0.0050432000309228895
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,16,0.004700800031423568
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,8,0.004892800003290176
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,4,0.004764800146222115
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,2,0.004796800017356872
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2,1,0.004691199958324432
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,128,0.008275199681520462
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,64,0.00647680014371872
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,32,0.004691199958324432
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,16,0.004716800153255462
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,8,0.0048191998153924945
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,4,0.004704000055789947
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,2,0.004806400090456009
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4,1,0.004806400090456009
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,128,0.008233600109815598
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,64,0.006438399851322174
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,32,0.0047680001705884935
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,16,0.004774399846792221
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,8,0.0047520000487566
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,4,0.004758400097489357
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,2,0.004639999940991402
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8,1,0.004758400097489357
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,128,0.008524800091981888
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,64,0.006227200105786324
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,32,0.005011200159788132
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,16,0.0047136001288890835
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,8,0.004764800146222115
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,4,0.004755200073122978
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,2,0.004687999933958053
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,16,1,0.004879999905824661
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,128,0.008627200126647949
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,64,0.006246399879455566
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,32,0.004691199958324432
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,16,0.0047807998955249785
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,8,0.004796800017356872
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,4,0.0046847999095916745
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,2,0.004771199822425842
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,32,1,0.004681599885225296
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,128,0.008463999629020691
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,64,0.006534399837255478
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,32,0.007238399982452392
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,16,0.005030399933457374
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,8,0.004806400090456009
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,4,0.004681599885225296
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,2,0.0047520000487566
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,48,1,0.004732799902558327
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,128,0.0069760002195835115
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,64,0.006390400230884552
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,32,0.006307200342416763
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,16,0.00605119988322258
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,8,0.004783999919891357
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,4,0.0047520000487566
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,2,0.004758400097489357
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,64,1,0.004796800017356872
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,128,0.008320000022649765
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,64,0.006825599819421768
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,32,0.006441599875688553
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,16,0.006172800064086914
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,8,0.005065599828958512
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,4,0.00493439994752407
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,2,0.0057183999568223955
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,80,1,0.004790399968624115
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,128,0.008665599673986436
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,64,0.00658240020275116
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,32,0.006335999816656113
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,16,0.005641600117087364
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,8,0.004729599878191948
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,4,0.004678399860858917
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,2,0.004742399975657463
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,96,1,0.0046720001846551895
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,128,0.008777599781751633
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,64,0.006707199662923813
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,32,0.006310400366783142
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,16,0.0062015999108552934
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,8,0.0047520000487566
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,4,0.0047807998955249785
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,2,0.0048351999372243885
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,128,1,0.00459199994802475
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,128,0.010489600151777268
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,64,0.008422400057315826
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,32,0.006656000018119812
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,16,0.00623680017888546
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,8,0.006207999959588051
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,4,0.005135999992489815
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,2,0.0051968000829219815
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,160,1,0.004985599964857102
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,128,0.010876800119876861
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,64,0.0086496002972126
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,32,0.006707199662923813
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,16,0.006351999938488007
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,8,0.005379199981689453
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,2,0.0049056001007556915
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,1,0.004758400097489357
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,128,0.01276479959487915
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,64,0.008540800213813782
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,32,0.006691200286149978
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,16,0.0065151996910572055
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,8,0.005606399849057198
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,4,0.0046431999653577805
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,2,0.004614400118589402
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,256,1,0.004575999826192856
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,128,0.014864000678062438
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,64,0.008841600269079208
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,32,0.006588800251483918
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,16,0.006457599997520447
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,8,0.00527999997138977
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,4,0.004828799888491631
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,2,0.004540799930691719
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,320,1,0.004534399881958961
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,128,0.01496960073709488
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,64,0.010633599758148194
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,32,0.008553600311279297
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,16,0.006563200056552887
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,8,0.006537599861621857
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,4,0.004748800024390221
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,2,0.0047807998955249785
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,384,1,0.00445760004222393
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,128,0.018857599794864656
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,64,0.01265919953584671
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,32,0.008755200356245042
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,16,0.006524799764156342
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,8,0.006447999924421311
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,4,0.006032000109553337
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,2,0.0074143998324871065
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,128,0.02842240035533905
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,64,0.016806399822235106
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,32,0.012918399274349212
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,16,0.008771199733018875
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,8,0.007308799773454666
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,4,0.006572800129652024
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,2,0.006444799900054932
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,768,1,0.006150399893522262
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,128,0.035955199599266054
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,64,0.0191551998257637
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,32,0.012937599420547485
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,16,0.010147199779748917
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,8,0.006908799707889557
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,4,0.006499200314283371
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,2,0.006387200206518173
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1024,1,0.005471999943256378
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,128,0.04870400130748749
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,64,0.02727360129356384
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,32,0.015177600085735321
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,16,0.012863999605178833
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,8,0.008748800307512284
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,4,0.0069983996450901035
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,2,0.006604799628257751
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1536,1,0.0060639999806880954
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,128,0.06250240206718445
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,64,0.03524479866027832
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,32,0.018831999599933626
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,16,0.012671999633312225
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,8,0.008793599903583527
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,4,0.00671359971165657
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,2,0.006505600363016129
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,2048,1,0.006095999851822853
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,128,0.08905919790267944
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,64,0.04769279956817627
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,32,0.026812800765037538
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,16,0.015228800475597382
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,8,0.01265919953584671
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,4,0.00894080027937889
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,2,0.006601600348949433
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,3072,1,0.00655680000782013
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,128,0.11620479822158813
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,64,0.06362879872322083
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,32,0.035395199060440065
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,16,0.018812799453735353
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,8,0.01266240030527115
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,4,0.008902399986982345
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,2,0.00687360018491745
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,4096,1,0.006374400109052658
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,128,0.17221120595932007
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,64,0.09023360013961793
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,32,0.05172799825668335
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,16,0.026953598856925963
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,8,0.014908799529075622
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,4,0.012572799623012543
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,2,0.008371199667453765
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,6144,1,0.006950400024652481
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,128,0.2222975969314575
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,64,0.11723519563674926
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,32,0.0652895987033844
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,16,0.035104000568389894
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,8,0.018742400407791137
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,4,0.012608000636100769
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,2,0.008953599631786347
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,8192,1,0.00687360018491745
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,1,64,0.006345599889755249
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,192,4,0.006224000081419945
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_pre,default,float16,512,1,0.004543999955058098
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,128,0.0065151996910572055
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,64,0.006128000095486641
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,32,0.006409599632024765
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,16,0.006233600154519081
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,8,0.006172800064086914
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,4,0.006224000081419945
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,2,0.006310400366783142
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1,1,0.004636799916625023
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,128,0.00655680000782013
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,64,0.00623680017888546
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,32,0.0064351998269557955
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,16,0.006351999938488007
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,8,0.006252799928188324
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,4,0.005164799839258194
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,2,0.004992000013589859
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2,1,0.0047807998955249785
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,128,0.0063391998410224915
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,64,0.006451199948787689
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,32,0.006521599739789963
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,16,0.00647680014371872
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,8,0.006259199976921081
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,4,0.005008000135421753
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,2,0.004937599971890449
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4,1,0.004681599885225296
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,128,0.006332799792289734
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,64,0.00623680017888546
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,32,0.006307200342416763
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,16,0.006438399851322174
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,8,0.006179200112819671
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,4,0.006080000102519989
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,2,0.006163199990987777
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8,1,0.004796800017356872
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,128,0.006723199784755707
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,64,0.006060799956321717
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,32,0.0062304001301527025
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,16,0.005913599953055381
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,8,0.006111999973654747
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,4,0.005171199887990951
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,2,0.005008000135421753
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16,1,0.00483199991285801
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,128,0.006790400296449661
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,64,0.006166400015354156
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,32,0.006294400244951248
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,16,0.00623680017888546
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,8,0.006387200206518173
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,4,0.006172800064086914
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,2,0.006268800050020218
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,32,1,0.004854400083422661
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,128,0.006752000004053116
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,64,0.006364800035953522
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,32,0.006195199862122536
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,16,0.006377600133419037
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,8,0.006361600011587143
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,4,0.0061471998691558834
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,2,0.006032000109553337
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,48,1,0.004812800139188766
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,128,0.006729599833488464
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,64,0.006486400216817856
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,32,0.006214400008320809
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,16,0.006179200112819671
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,8,0.00613120011985302
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,4,0.005427199974656105
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,2,0.0047391999512910845
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,64,1,0.004720000177621841
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,128,0.006905599683523178
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,64,0.006511999666690827
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,32,0.0062015999108552934
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,16,0.006281600147485733
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,8,0.00639680027961731
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,4,0.00634239986538887
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,2,0.005289600044488907
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,80,1,0.004800000041723251
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,128,0.007433599978685379
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,64,0.006326399743556976
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,32,0.006384000182151794
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,16,0.0061055999249219894
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,8,0.006335999816656113
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,2,0.006179200112819671
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,1,0.005052800104022026
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,128,0.008377599716186523
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,64,0.006569600105285645
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,32,0.0064191997051239015
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,16,0.006470400094985962
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,8,0.006415999680757523
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,4,0.006374400109052658
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,2,0.006246399879455566
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,128,1,0.005023999884724617
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,128,0.00843840017914772
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,64,0.006665600091218948
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,32,0.006304000318050384
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,16,0.0063231997191905975
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,8,0.006304000318050384
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,4,0.006188800185918808
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,2,0.005612799897789955
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,160,1,0.005257600173354149
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,128,0.008716800063848496
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,64,0.006934399902820587
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,32,0.006425599753856659
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,96,4,0.0062720000743865965
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,8,0.00623680017888546
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,4,0.006534399837255478
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,2,0.006278400123119354
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,1,0.006239999830722809
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,128,0.009200000017881394
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,64,0.007119999825954437
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,32,0.0066143997013568875
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,16,0.0065600000321865085
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,8,0.006198399886488915
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,4,0.00621119998395443
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,2,0.006364800035953522
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,256,1,0.005859199911355972
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,128,0.010780800133943558
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,64,0.008761599659919739
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,32,0.006889600306749344
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,16,0.0064191997051239015
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,8,0.006304000318050384
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,4,0.00613120011985302
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,2,0.006335999816656113
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,320,1,0.005475199967622757
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,128,0.01112319976091385
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,64,0.00880960002541542
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,32,0.006825599819421768
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,16,0.00658240020275116
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,8,0.006473600119352341
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,4,0.00642239972949028
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,2,0.006377600133419037
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,384,1,0.006361600011587143
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,128,0.015187199413776397
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,64,0.00896959975361824
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,32,0.007014399766921997
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,16,0.006521599739789963
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,8,0.006470400094985962
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,4,0.0063231997191905975
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,2,0.00605119988322258
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,512,1,0.006326399743556976
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,128,0.027046400308609008
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,64,0.011036799848079681
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,32,0.00870719999074936
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,16,0.006553599983453751
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,8,0.006569600105285645
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,4,0.006406400352716446
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,2,0.006415999680757523
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,768,1,0.006217600032687187
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,128,0.0333759993314743
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,64,0.013264000415802002
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,32,0.00886079967021942
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,16,0.006844799965620041
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,8,0.0064351998269557955
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,4,0.006204799935221672
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,2,0.006284800171852112
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1024,1,0.006169600039720535
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,128,0.04753600060939789
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,64,0.025164800882339477
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,32,0.010921599715948105
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,16,0.008748800307512284
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,8,0.006774400174617767
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,4,0.006374400109052658
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,2,0.006566400080919266
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,1536,1,0.006358399987220764
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,128,0.06395840048789977
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,64,0.031430399417877196
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,32,0.012771199643611907
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,16,0.009068799763917923
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,8,0.006809599697589874
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,4,0.006492800265550614
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,2,0.006329599767923355
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,2048,1,0.006387200206518173
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,128,0.08517760038375854
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,64,0.0461760014295578
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,32,0.026924800872802735
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,16,0.010815999656915664
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,8,0.008668799698352814
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,4,0.006825599819421768
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,2,0.006355199962854385
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,3072,1,0.00628800019621849
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,128,0.11974079608917236
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,64,0.0637440025806427
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,32,0.03306879997253418
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,16,0.012966400384902954
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,8,0.008883199840784072
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,4,0.00687360018491745
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,2,0.006294400244951248
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,4096,1,0.006313599646091461
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,128,0.16316800117492675
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,64,0.0865343987941742
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,32,0.05089600086212158
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,16,0.024710400402545928
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,8,0.01080000028014183
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,4,0.008511999994516373
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,2,0.006771200150251388
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,6144,1,0.006255999952554703
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,128,0.22912321090698243
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,64,0.11945279836654663
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,32,0.06649600267410279
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,16,0.03115200102329254
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,8,0.013305599987506866
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,4,0.008806400001049042
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,2,0.006908799707889557
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,8192,1,0.006441599875688553
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,128,0.33618879318237305
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,64,0.17414720058441163
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,32,0.08910719752311706
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,16,0.04620159864425659
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,8,0.023772799968719484
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,1,0.007119999825954437
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,128,0.4441728115081787
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,64,0.23463358879089355
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,32,0.1308608055114746
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,16,0.063372802734375
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,8,0.03162240087985992
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,4,0.013235199451446533
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,2,0.00910080000758171
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,16384,1,0.006860800087451935
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,2,0.008691199868917466
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,12288,4,0.010956799983978272
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,128,0.5669983863830567
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,16,0.0750432014465332
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,32,0.15121279954910277
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,64,0.28519999980926514
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,8,0.040278398990631105
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,1,0.00822720006108284
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,4,0.01815039962530136
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,20480,2,0.008720000088214875
TRTLLM,1.2.0rc5,NVIDIA GB200,mla_gen_post,default,float16,192,16,0.006259199976921081
