framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,128,0.009062399715185165
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,64,0.007065600156784058
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,32,0.0069983996450901035
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,8,0.00735040009021759
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,16,0.0076672002673149105
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,4,0.008342400193214417
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,2,0.007983999699354172
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,64,0.007004799693822861
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,32,0.007542400062084198
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,16,0.006819199770689011
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1,1,0.006304000318050384
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,128,0.009113600105047226
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,8,0.007868800312280655
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,4,0.006819199770689011
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,2,0.006275200098752975
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,128,0.009433600306510925
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2,1,0.007923199981451034
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,64,0.008240000158548356
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,32,0.0065311998128890995
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,16,0.00719040036201477
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,8,0.007372800260782242
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,4,0.007078400254249573
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,2,0.007254400104284286
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4,1,0.007353600114583969
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,128,0.00920960009098053
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,64,0.007152000069618225
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,32,0.007065600156784058
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,16,0.007865600287914276
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,4,0.006972800195217133
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,8,0.007779199630022049
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,1,0.007686399668455124
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8,2,0.006876800209283829
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,128,0.009388799965381622
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,64,0.007664000242948532
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,32,0.007766400277614593
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,16,0.007344000041484833
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,8,0.0069472000002861025
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,4,0.007699199765920639
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,2,0.010051199793815612
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,16,1,0.006428799778223038
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,128,0.009497600048780442
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,64,0.008387199789285659
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,32,0.007475200295448303
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,16,0.007142399996519088
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,8,0.008656000345945358
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,4,0.0073183998465538025
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,2,0.007654400169849395
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,32,1,0.006454399973154068
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,128,0.010742399841547012
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,32,0.008022399991750718
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,64,0.009539200365543366
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,8,0.009356799721717834
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,16,0.006707199662923813
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,4,0.007999999821186066
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,1,0.0071712002158164975
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,48,2,0.006841599941253662
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,128,0.009151999652385712
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,64,0.009523200243711472
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,32,0.0065311998128890995
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,16,0.007561600208282471
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,8,0.007097599655389785
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,4,0.008310399949550629
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,2,0.0070271998643875126
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,64,1,0.008947200328111648
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,128,0.009040000289678574
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,32,0.007788799703121185
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,64,0.007596799731254577
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,8,0.0062720000743865965
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,16,0.009801600128412247
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,4,0.007196799665689468
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,2,0.007411199808120728
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,80,1,0.0070720002055168155
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,128,0.009424000233411788
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,64,0.007676800340414047
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,32,0.00982080027461052
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,16,0.007340800017118454
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,8,0.007628799974918365
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,4,0.0069760002195835115
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,2,0.007119999825954437
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,96,1,0.008841600269079208
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,64,0.008345600217580795
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,128,0.009353599697351455
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,32,0.008767999708652496
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,16,0.00679360032081604
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,8,0.0068800002336502075
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,4,0.00838399976491928
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,2,0.007660800218582153
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,128,0.011526399850845337
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,128,1,0.006649599969387054
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,32,0.009916800260543823
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,64,0.00886079967021942
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,16,0.0074432000517845156
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,8,0.007254400104284286
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,4,0.00700799971818924
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,2,0.0071263998746871945
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,160,1,0.006911999732255936
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,128,0.013734400272369385
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,64,0.009216000139713288
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,32,0.008294399827718735
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,16,0.007078400254249573
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,4,0.007887999713420867
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,8,0.0071263998746871945
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,2,0.008303999900817871
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,192,1,0.007664000242948532
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,128,0.013532799482345582
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,64,0.009654399752616883
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,16,0.007440000027418137
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,32,0.007817599922418594
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,8,0.007340800017118454
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,4,0.007119999825954437
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,2,0.008320000022649765
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,256,1,0.008303999900817871
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,128,0.01560640037059784
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,64,0.01141119971871376
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,32,0.009171199798583985
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,16,0.008339200168848038
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,8,0.007334399968385696
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,4,0.0070592001080513
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,2,0.008233600109815598
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,128,0.017452800273895265
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,320,1,0.006943999975919724
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,64,0.011664000153541566
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,32,0.00944959968328476
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,16,0.007708799839019775
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,8,0.007980799674987793
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,2,0.007056000083684922
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,4,0.007398399710655213
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,384,1,0.007478400319814682
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,128,0.02165440022945404
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,64,0.015347200632095336
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,32,0.009481599926948548
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,16,0.00907519981265068
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,4,0.008870399743318557
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,8,0.00904960036277771
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,2,0.007100799679756164
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,512,1,0.007536000013351441
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,128,0.030393600463867188
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,64,0.01775680035352707
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,32,0.013334399461746216
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,16,0.0095040000975132
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,8,0.00793280005455017
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,4,0.008150400221347808
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,2,0.007193599641323089
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,768,1,0.0076223999261856076
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,128,0.03668160140514374
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,64,0.021657599508762358
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,32,0.013571199774742127
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,16,0.011184000223875046
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,8,0.007609599828720092
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,2,0.007327999919652939
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,4,0.006828799843788147
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1024,1,0.007571200281381607
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,128,0.049711999297142026
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,64,0.028252801299095152
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,16,0.013436800241470337
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,32,0.01738879978656769
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,8,0.009344000369310379
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,128,0.06383360028266907
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,4,0.007456000149250031
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,1,0.008294399827718735
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,64,0.03526720106601715
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,32,0.020028799772262573
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,16,0.013507199287414551
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,8,0.009411200135946273
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,4,0.007846400141716003
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,2,0.007840000092983246
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,2048,1,0.008032000064849854
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,128,0.08956159949302674
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,64,0.04932479858398438
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,32,0.028387200832366944
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,16,0.01749120056629181
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,8,0.013344000279903411
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,4,0.009408000111579894
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,2,0.007977599650621415
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,128,0.11688319444656373
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,3072,1,0.007791999727487564
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,64,0.06242560148239136
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,32,0.03546879887580871
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,16,0.020259200036525725
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,4,0.00968639999628067
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,8,0.013504000008106231
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,2,0.00793280005455017
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,4096,1,0.008271999657154083
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,128,0.17059839963912965
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,64,0.09048640131950378
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,32,0.050108802318572995
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,16,0.028889599442481994
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,8,0.017459200322628023
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,4,0.013033600151538849
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,2,0.009487999975681305
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,128,0.22312960624694825
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,6144,1,0.00761599987745285
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,64,0.1160863995552063
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,32,0.06557760238647461
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,16,0.03630079925060272
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,8,0.019411200284957887
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,4,0.013459199666976928
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,2,0.010127999633550645
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,8192,1,0.00764160007238388
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_pre,default,float16,1536,2,0.00764160007238388
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,128,0.007369600236415863
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,32,0.007142399996519088
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,8,0.006940799951553345
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,64,0.007407999783754349
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,16,0.007372800260782242
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,4,0.006844799965620041
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,2,0.0088128000497818
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1,1,0.006857600063085556
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,128,0.00713919997215271
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,32,0.007724799960851669
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,16,0.007254400104284286
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,8,0.007187200337648391
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,64,0.009142400324344635
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,4,0.007481600344181061
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,2,0.00684799998998642
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2,1,0.006761600077152252
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,128,0.011475200206041336
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,64,0.006911999732255936
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,32,0.006886400282382965
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,16,0.009148799628019334
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,2,0.006598400324583054
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,1,0.006822399795055389
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,128,0.007248000055551529
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,64,0.007036799937486649
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,32,0.007193599641323089
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,16,0.007068800181150437
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,8,0.007231999933719635
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,4,0.007523199915885926
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,2,0.008054400235414505
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8,1,0.0084927998483181
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,64,0.007100799679756164
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,128,0.007273600250482559
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,32,0.007161600142717361
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,16,0.00650240033864975
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,8,0.007478400319814682
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,4,0.00713919997215271
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,2,0.007068800181150437
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16,1,0.006854400038719177
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,128,0.007702399790287018
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,64,0.007231999933719635
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,32,0.007116799801588058
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,16,0.008579199761152267
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,8,0.00719040036201477
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,4,0.007215999811887741
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,2,0.007142399996519088
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,32,1,0.007772800326347351
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,128,0.00740479975938797
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,64,0.0076384000480175015
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,32,0.00759039968252182
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,16,0.007135999947786331
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,8,0.0069311998784542085
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,4,0.007276800274848938
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,2,0.007014399766921997
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,48,1,0.00713919997215271
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,128,0.008377599716186523
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,64,0.007084800302982331
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,32,0.006735999882221222
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,16,0.0070592001080513
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,8,0.00732479989528656
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,4,0.0074432000517845156
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,2,0.006732799857854843
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,64,1,0.008163200318813324
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,128,0.008921600133180618
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,64,0.007558400183916092
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,32,0.008079999685287475
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,16,0.007334399968385696
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,8,0.007379200309514999
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,4,0.008012799918651581
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,2,0.007478400319814682
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,80,1,0.007644800096750259
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,128,0.008825600147247314
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,64,0.007401599735021591
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,32,0.007238399982452392
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,16,0.007103999704122543
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,8,0.006857600063085556
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,4,0.0075392000377178196
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,2,0.007391999661922455
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,96,1,0.007382400333881378
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,128,0.009046400338411332
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,64,0.008028800040483475
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,32,0.007273600250482559
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,16,0.007868800312280655
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,8,0.008044800162315369
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,4,0.006969600170850754
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,2,0.007081600278615952
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,128,1,0.007648000121116638
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,128,0.009046400338411332
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,64,0.007529599964618683
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,32,0.007075200229883194
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,16,0.007481600344181061
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,8,0.00676800012588501
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,4,0.0072223998606204985
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,2,0.006579200178384781
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,160,1,0.0100832000374794
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,128,0.00952640026807785
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,64,0.00902400016784668
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,32,0.007715199887752533
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,16,0.00740479975938797
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,8,0.007516799867153168
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,4,0.00740479975938797
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,2,0.008656000345945358
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,192,1,0.006694400310516357
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,128,0.009891200065612792
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,64,0.009276799857616425
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,32,0.007011199742555619
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,16,0.007497599720954895
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,8,0.007248000055551529
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,4,0.00735040009021759
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,2,0.007286400347948074
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,256,1,0.007456000149250031
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,128,0.011372800171375274
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,64,0.009232000261545182
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,32,0.0075552001595497135
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,16,0.007452800124883652
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,8,0.007478400319814682
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,4,0.007516799867153168
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,2,0.006969600170850754
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,320,1,0.007004799693822861
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,128,0.011734399944543839
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,64,0.009785600006580353
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,32,0.007760000228881836
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,16,0.006883200258016586
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,8,0.007177600264549255
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,4,0.007187200337648391
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,2,0.0070592001080513
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,384,1,0.0074143998324871065
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,128,0.01610880047082901
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,64,0.010966400057077408
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,32,0.007740800082683563
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,8,0.007516799867153168
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,16,0.0073472000658512115
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,4,0.006937599927186966
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,2,0.007388799637556076
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,512,1,0.007113599777221679
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,128,0.029631999135017396
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,64,0.01159999966621399
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,32,0.009369599819183349
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,16,0.007651200145483017
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,8,0.00735040009021759
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,4,0.007020799815654755
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,2,0.009728000313043595
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,768,1,0.008287999778985977
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,128,0.03567039966583252
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,64,0.01530880033969879
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,32,0.010835199803113937
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,16,0.007545600086450577
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,8,0.008076799660921096
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,4,0.007273600250482559
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,2,0.006943999975919724
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1024,1,0.007859200239181519
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,128,0.049395200610160825
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,64,0.028857600688934327
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,32,0.012988799810409546
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,16,0.00912960022687912
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,8,0.007513599842786789
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,4,0.007248000055551529
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,2,0.0075935997068881985
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,1536,1,0.008214399963617326
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,128,0.06403520107269287
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,64,0.03549439907073974
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,32,0.015311999619007111
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,16,0.009657599776983262
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,8,0.010236799716949463
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,4,0.007199999690055847
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,2,0.007091200351715088
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,2048,1,0.0070271998643875126
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,128,0.08862079977989197
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,64,0.05033919811248779
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,32,0.029583999514579774
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,16,0.011299200356006622
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,8,0.009296000003814697
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,4,0.007654400169849395
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,2,0.007977599650621415
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,3072,1,0.00769599974155426
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,128,0.117084801197052
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,64,0.06396160125732422
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,32,0.03338240087032318
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,16,0.013657599687576294
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,8,0.011116799712181092
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,4,0.007292799651622772
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,2,0.007184000313282013
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4096,1,0.008073599636554718
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,128,0.17223680019378662
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,64,0.08840320110321045
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,16,0.027584001421928406
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,32,0.04960319995880127
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,8,0.011462400108575821
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,4,0.009510400146245957
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,2,0.007264000177383423
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,6144,1,0.007321599870920181
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,128,0.2237407922744751
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,64,0.1183135986328125
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,32,0.06724479794502258
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,16,0.03370560109615326
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,8,0.014627200365066529
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,4,0.011155200004577637
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,2,0.007599999755620956
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,8192,1,0.007369600236415863
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,128,0.33044478893280027
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,64,0.17448960542678832
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,32,0.09294400215148926
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,16,0.04734399914741516
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,8,0.027619200944900512
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,4,0.011695999652147293
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,2,0.009273599833250046
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,12288,1,0.007542400062084198
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,128,0.425164794921875
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,64,0.23101439476013183
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,32,0.12100160121917725
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,16,0.06251199841499329
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,8,0.03358719944953918
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,4,0.015369600057601929
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,2,0.009545599669218063
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,16384,1,0.00894080027937889
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,128,0.5328896045684814
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,64,0.28538880348205564
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,32,0.14111360311508178
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,16,0.07684159874916077
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,8,0.04092159867286682
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,4,0.019411200284957887
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,2,0.011609599739313126
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,20480,1,0.009596800059080124
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,8,0.007903999835252761
TRTLLM,1.0.0rc6,NVIDIA GB200,mla_gen_post,default,float16,4,4,0.006703999638557434
