framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,128,0.006662400066852569
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,16,0.004521600157022476
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,8,0.004508800059556961
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,4,0.00448639988899231
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,2,0.004492799937725067
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,1,0.0044351998716592785
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,128,0.006569600105285645
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,64,0.004822399839758873
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,32,0.004441599920392036
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,16,0.004611200094223023
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,8,0.004687999933958053
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,4,0.004441599920392036
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,2,0.004700800031423568
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,1,0.00453759990632534
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,128,0.0068800002336502075
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,64,0.006508799642324448
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,32,0.006115199998021126
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,16,0.005081599950790406
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,8,0.004745600000023842
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,4,0.004623999819159508
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,2,0.004694399982690811
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,1,0.004704000055789947
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,128,0.006815999746322632
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,64,0.006326399743556976
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,32,0.004704000055789947
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,16,0.0046847999095916745
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,8,0.0047520000487566
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,4,0.0048191998153924945
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,2,0.0047231998294591905
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,64,0.004652800038456917
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,32,0.004579199850559235
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,1,0.0048096001148223875
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,64,0.006384000182151794
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,32,0.004902400076389313
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,16,0.004675199836492538
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,8,0.0047775998711586
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,2,0.004665600135922432
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,1,0.004729599878191948
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,128,0.008320000022649765
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,64,0.006332799792289734
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,32,0.0063231997191905975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,16,0.004659200087189674
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,8,0.004992000013589859
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,4,0.004822399839758873
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,2,0.004944000020623207
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,128,0.008463999629020691
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,64,0.00639680027961731
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,128,0.006598400324583054
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,16,0.004646399989724159
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,4,0.004492799937725067
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,4,0.004646399989724159
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,2,0.004476799815893173
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,1,0.0045311998575925825
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,128,0.006537599861621857
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,64,0.006467200070619583
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,32,0.004524800181388855
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,16,0.004569600149989128
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,8,0.004646399989724159
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,4,0.004521600157022476
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,2,0.0045311998575925825
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,1,0.004915200173854828
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,128,0.008611200004816055
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,64,0.006553599983453751
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,32,0.006403200328350067
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,16,0.0046016000211238865
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,8,0.004534399881958961
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,4,0.004630399867892265
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,2,0.004636799916625023
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,1,0.004527999833226204
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,128,0.008495999872684479
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,64,0.006601600348949433
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,32,0.006102399900555611
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,16,0.004476799815893173
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,8,0.004793599992990494
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,4,0.004428799822926521
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,2,0.004662400111556053
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,1,0.004604800045490265
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,128,0.008611200004816055
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,32,0.006124800071120262
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,32,0.006403200328350067
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,8,0.0045056000351905824
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,8,0.004764800146222115
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,4,0.004540799930691719
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,1,0.005071999877691269
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,1,0.004879999905824661
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,128,0.010691200196743012
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,64,0.00759039968252182
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,32,0.0065151996910572055
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,16,0.006255999952554703
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,8,0.004924799874424934
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,4,0.004783999919891357
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,2,0.00522879995405674
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,1,0.0048191998153924945
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,64,0.006483200192451477
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,128,0.010841599851846694
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,64,0.008451200276613235
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,16,0.006070400029420853
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,8,0.0053311999887228016
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,4,0.0047775998711586
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,2,0.005219199880957604
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,32,0.006672000139951706
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,16,0.005689600110054016
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,128,0.011673600226640702
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,64,0.008601599931716919
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,32,0.006553599983453751
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,16,0.0064800001680850984
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,8,0.004758400097489357
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,4,0.004502400010824204
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,2,0.0044895999133586885
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,1,0.005910399928689003
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,128,0.013302400708198547
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,64,0.009148799628019334
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,32,0.006537599861621857
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,16,0.006409599632024765
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,8,0.00621119998395443
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,4,0.004588799923658371
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,1,0.005036799982190132
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,1,0.005094400048255921
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,128,0.015167999267578124
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,64,0.0106175996363163
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,32,0.0084927998483181
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,16,0.006889600306749344
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,4,0.005513599887490272
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,2,0.004646399989724159
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,1,0.004678399860858917
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,128,0.018812799453735353
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,64,0.012700800597667695
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,32,0.008511999994516373
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,16,0.006732799857854843
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,8,0.00631679967045784
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,4,0.005471999943256378
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,2,0.004582399874925614
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,1,0.004575999826192856
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,128,0.027849599719047546
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,64,0.014774399995803832
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,8,0.006384000182151794
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,16,0.008399999886751174
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,8,0.006969600170850754
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,4,0.006505600363016129
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,2,0.0065151996910572055
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,1,0.005446400120854377
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,128,0.0356799989938736
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,64,0.018854400515556334
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,32,0.010956799983978272
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,2,0.004569600149989128
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,8,0.0066431999206542965
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,2,0.00469760000705719
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,2,0.006099199876189232
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,1,0.005513599887490272
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,32,0.010902400314807891
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,64,0.026873600482940675
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,32,0.014752000570297241
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,16,0.010815999656915664
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,8,0.008515200018882752
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,4,0.0065760001540184024
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,2,0.006403200328350067
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,1,0.005238400027155876
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,128,0.06190720200538635
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,64,0.03468480110168457
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,32,0.017177599668502807
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,16,0.012656000256538392
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,8,0.00859839990735054
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,4,0.006672000139951706
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,2,0.006579200178384781
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,128,0.04782719910144806
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,128,0.08809599876403809
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,64,0.0474047988653183
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,32,0.02720000147819519
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,16,0.014975999295711518
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,8,0.010755199939012527
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,4,0.00854720026254654
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,2,0.0068000003695487974
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,1,0.006438399851322174
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,16,0.008883199840784072
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,64,0.062249600887298584
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,4,0.00650240033864975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,16,0.016995200514793397
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,8,0.012591999769210816
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,4,0.008627200126647949
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,2,0.006672000139951706
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,1,0.006377600133419037
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,128,0.1658079981803894
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,64,0.08897600173950196
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,32,0.047788798809051514
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,16,0.02688319981098175
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,8,0.014771200716495514
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,4,0.010812799632549285
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,128,0.11599040031433105
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,1,0.0068800002336502075
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,32,0.035043200850486754
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,128,0.2192863941192627
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,32,0.06454399824142457
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,16,0.03353599905967712
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,8,0.016947199404239655
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,4,0.012646399438381195
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,2,0.008879999816417693
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,1,0.00671359971165657
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,2,0.008550400286912918
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,64,0.11900800466537476
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,1,0.005356800183653832
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,128,0.006431999802589417
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,64,0.00650240033864975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,32,0.006441599875688553
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,16,0.00623680017888546
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,8,0.00631679967045784
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,4,0.005641600117087364
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,2,0.004892800003290176
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,1,0.004812800139188766
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,128,0.006361600011587143
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,64,0.006639999896287918
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,32,0.006326399743556976
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,16,0.006294400244951248
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,8,0.006412799656391144
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,4,0.0062015999108552934
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,2,0.006207999959588051
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,1,0.005932800099253654
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,128,0.006172800064086914
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,64,0.006281600147485733
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,32,0.006415999680757523
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,16,0.006255999952554703
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,8,0.006371200084686279
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,4,0.006060799956321717
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,2,0.006070400029420853
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,1,0.006012799963355064
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,128,0.006268800050020218
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,64,0.005497600138187409
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,32,0.0063680000603199005
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,16,0.0063231997191905975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,8,0.006335999816656113
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,4,0.005635200068354607
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,2,0.004982399940490723
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,1,0.0051968000829219815
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,128,0.006876800209283829
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,64,0.006124800071120262
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,32,0.00623680017888546
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,16,0.0061055999249219894
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,8,0.006332799792289734
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,4,0.0061760000884532925
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,2,0.006198399886488915
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,1,0.0063231997191905975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,128,0.006601600348949433
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,64,0.0063231997191905975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,32,0.006319999694824219
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,16,0.00615679994225502
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,8,0.005999999865889549
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,4,0.005667199939489364
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,2,0.005087999999523163
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,1,0.004886399954557419
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,128,0.006649599969387054
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,64,0.006457599997520447
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,32,0.006393600255250931
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,16,0.006111999973654747
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,8,0.00639680027961731
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,4,0.0052352000027894975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,2,0.0049183998256921765
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,1,0.004889599978923798
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,128,0.0068351998925209045
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,16,0.006374400109052658
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,8,0.006335999816656113
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,4,0.006118400022387505
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,2,0.006025600060820579
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,1,0.006601600348949433
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,128,0.00679360032081604
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,64,0.00639680027961731
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,32,0.006239999830722809
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,16,0.006415999680757523
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,8,0.006185600161552429
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,4,0.005859199911355972
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,2,0.005126399919390678
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,1,0.005100800096988678
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,128,0.006803199648857117
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,64,0.006393600255250931
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,32,0.006294400244951248
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,16,0.006400000303983688
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,8,0.006153599917888641
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,4,0.006252799928188324
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,2,0.00613120011985302
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,1,0.006070400029420853
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,128,0.008374399691820144
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,64,0.0063680000603199005
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,32,0.006371200084686279
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,16,0.00631679967045784
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,8,0.006326399743556976
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,4,0.0064351998269557955
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,2,0.0061471998691558834
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,1,0.006217600032687187
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,128,0.008636800199747085
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,64,0.006726399809122085
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,32,0.006319999694824219
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,16,0.006179200112819671
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,8,0.00618240013718605
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,4,0.005686400085687637
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,2,0.004995200037956238
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,64,0.00642239972949028
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,32,0.006403200328350067
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,64,0.006828799843788147
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,32,0.006313599646091461
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,16,0.006259199976921081
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,8,0.006335999816656113
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,4,0.006022400036454201
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,1,0.004800000041723251
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,128,0.008604799956083297
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,2,0.004982399940490723
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,1,0.004569600149989128
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,64,0.006566400080919266
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,32,0.006483200192451477
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,16,0.00642239972949028
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,8,0.004556800052523613
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,4,0.0046016000211238865
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,2,0.004828799888491631
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,1,0.0045056000351905824
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,128,0.01071999967098236
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,32,0.006579200178384781
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,16,0.006345599889755249
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,128,0.008796799927949905
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,4,0.006326399743556976
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,2,0.0063231997191905975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,1,0.006198399886488915
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,128,0.010790400207042694
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,64,0.008604799956083297
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,32,0.006703999638557434
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,16,0.00623680017888546
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,8,0.0062015999108552934
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,4,0.006195199862122536
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,1,0.004806400090456009
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,128,0.015270400047302245
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,64,0.008767999708652496
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,32,0.006809599697589874
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,16,0.006249599903821945
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,8,0.0064640000462532045
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,4,0.006428799778223038
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,2,0.006198399886488915
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,1,0.0048351999372243885
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,128,0.026982399821281432
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,64,0.010623999685049058
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,32,0.008528000116348267
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,16,0.006652799993753433
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,8,0.0064800001680850984
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,4,0.006387200206518173
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,2,0.006371200084686279
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,1,0.0068063996732234955
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,128,0.034108799695968625
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,64,0.01324159950017929
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,32,0.00920960009098053
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,16,0.0072672002017498015
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,8,0.006752000004053116
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,4,0.006598400324583054
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,64,0.006787200272083282
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,1,0.0047775998711586
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,128,0.04755200147628784
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,64,0.02502720057964325
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,32,0.010911999642848969
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,16,0.008591999858617782
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,8,0.00663359984755516
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,4,0.006364800035953522
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,2,0.00623680017888546
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,1,0.006524799764156342
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,128,0.06254720091819763
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,64,0.033232000470161435
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,32,0.012761600315570831
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,16,0.00862400010228157
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,8,0.006703999638557434
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,4,0.006451199948787689
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,2,0.006294400244951248
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,1,0.006387200206518173
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,128,0.08470720052719116
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,64,0.04808320105075836
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,32,0.025119999051094057
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,16,0.010659199953079224
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,8,0.008633600175380706
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,4,0.006665600091218948
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,2,0.006275200098752975
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,1,0.006179200112819671
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,128,0.1176192045211792
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,64,0.06287680268287658
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,32,0.03315840065479279
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,16,0.012649600207805634
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,8,0.008575999736785888
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,4,0.0066143997013568875
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,2,0.006380800157785416
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,1,0.006489600241184235
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,128,0.16454399824142457
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,64,0.08676159977912903
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,32,0.04601280093193054
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,16,0.02451840043067932
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,8,0.010684800148010255
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,4,0.008563199639320373
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,2,0.006534399837255478
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,1,0.006489600241184235
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,128,0.2233344078063965
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,64,0.1190335988998413
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,32,0.06380159854888916
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,16,0.031065601110458373
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,8,0.012720000743865967
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,4,0.00854720026254654
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,2,0.006598400324583054
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,1,0.006425599753856659
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,8,0.006332799792289734
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,128,0.33270080089569093
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,32,0.08914560079574585
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,2,0.006470400094985962
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,16,0.04595839977264404
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,8,0.024134400486946105
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,4,0.010911999642848969
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,2,0.008764799684286118
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,1,0.007123199850320816
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,128,0.4482016086578369
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,64,0.2298207998275757
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,32,0.1239967942237854
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,16,0.060227197408676145
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,8,0.031404799222946166
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,4,0.013116799294948578
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,2,0.009011200070381165
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,1,0.006995200365781784
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,2,0.006329599767923355
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,64,0.28181118965148927
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,32,0.15114879608154297
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,16,0.07476800084114074
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,8,0.03765439987182617
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,4,0.016927999258041383
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,2,0.008767999708652496
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,1,0.008591999858617782
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,128,0.5605567932128906
TRTLLM,1.2.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,64,0.17023359537124633
