framework,version,device,op_name,kernel_source,moe_dtype,num_tokens,hidden_size,topk,num_experts,moe_ep_size,distribution,latency
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,4,balanced,0.06102080084383488
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,4,balanced,0.05054400004446506
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,4,balanced,0.07497600093483925
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,4,balanced,0.05843839980661869
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,4,balanced,0.04726719968020916
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,4,balanced,0.07874239906668663
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,4,balanced,0.06161920018494129
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,4,balanced,0.051254399493336676
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,4,balanced,0.0769152007997036
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,4,balanced,0.07670399844646454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,4,balanced,0.05836479924619198
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,4,balanced,0.04762879982590675
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,4,balanced,0.07557120025157929
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,4,balanced,0.06228479966521263
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,4,balanced,0.04920640029013157
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,4,balanced,0.07535679936408997
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,4,balanced,0.05746559947729111
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,4,balanced,0.04635839983820915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,4,balanced,0.0730432003736496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,4,balanced,0.07397119998931885
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,4,balanced,0.056086399406194684
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,4,balanced,0.045107199996709826
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,4,balanced,0.07555200159549713
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,4,balanced,0.056729599460959436
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,4,balanced,0.04312639981508255
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,4,balanced,0.07605760097503662
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,4,balanced,0.05757759995758534
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,4,balanced,0.047817600145936015
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,4,balanced,0.07458240017294884
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,4,balanced,0.07496960014104843
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,4,balanced,0.05825279988348484
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,4,balanced,0.04712320007383823
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,4,balanced,0.07536640092730522
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,4,balanced,0.05482240058481693
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,4,balanced,0.04493759945034981
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,4,balanced,0.07255359888076782
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,4,balanced,0.05637120045721531
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,4,balanced,0.04583680033683777
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,4,balanced,0.0707104004919529
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,4,balanced,0.07240320071578026
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,4,balanced,0.05040000006556511
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,4,balanced,0.046281599625945094
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,4,balanced,0.07463999986648559
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,4,balanced,0.0556063998490572
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,4,balanced,0.043622399866580966
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,4,balanced,0.07218879908323288
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,4,balanced,0.05126079991459846
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,4,balanced,0.045702400431036946
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,4,balanced,0.07168320044875146
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,4,balanced,0.07384000048041343
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,4,balanced,0.05517119951546192
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,4,balanced,0.043859200179576875
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,4,balanced,0.07165760099887848
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,4,balanced,0.05319040007889271
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,4,balanced,0.04251200035214424
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,4,balanced,0.06949120089411735
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,4,balanced,0.05454399958252907
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,4,balanced,0.04650239981710911
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,4,balanced,0.07259199991822243
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,4,balanced,0.07135039940476418
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,4,balanced,0.05456320010125637
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,4,balanced,0.04258880019187927
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,4,balanced,0.06969599984586239
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,4,balanced,0.05400639958679676
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,4,balanced,0.04457919932901859
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,4,balanced,0.07099519930779934
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,4,balanced,0.050889599695801736
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,4,balanced,0.046636800095438954
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,4,balanced,0.06892160065472126
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,4,balanced,0.07071359977126121
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,4,balanced,0.054163200035691264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,4,balanced,0.04409919939935207
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,4,balanced,0.07102400064468384
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,4,balanced,0.052134400606155394
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,4,balanced,0.04472639970481396
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,4,balanced,0.06783680021762847
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,4,balanced,0.05366399995982647
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,4,balanced,0.04569280035793781
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,4,balanced,0.06874239966273307
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,4,balanced,0.06989120095968246
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,4,balanced,0.05292800031602383
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,4,balanced,0.04447040036320686
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,4,balanced,0.06843520067632199
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,4,balanced,0.0542239997535944
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,4,balanced,0.041430399566888806
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,4,balanced,0.07022719904780388
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,4,balanced,0.05508800074458122
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,4,balanced,0.04491840004920959
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,4,balanced,0.07439680024981499
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,4,balanced,0.07186559960246086
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,4,balanced,0.0534752007573843
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,4,balanced,0.04287359975278378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,4,balanced,0.06670400053262711
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,4,balanced,0.050889600068330765
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,4,balanced,0.04454080015420914
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,4,balanced,0.06874879971146583
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,4,balanced,0.05118079967796803
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,4,balanced,0.04415040016174317
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,4,balanced,0.06495999954640866
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,4,balanced,0.07318399995565414
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,4,balanced,0.05244480036199093
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,4,balanced,0.04235200062394142
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,4,balanced,0.07200959995388985
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,4,balanced,0.05080000050365925
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,4,balanced,0.039059200510382655
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,4,balanced,0.06828800067305565
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,4,balanced,0.05384000018239021
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,4,balanced,0.04483839981257916
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,4,balanced,0.0684736005961895
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,4,balanced,0.07072000056505204
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,4,balanced,0.053561600297689436
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,4,balanced,0.04325439967215061
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,4,balanced,0.07009920105338097
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,4,balanced,0.04978879988193512
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,4,balanced,0.0416799996048212
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,4,balanced,0.06956799998879433
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,4,balanced,0.05334080047905445
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,4,balanced,0.04374399967491627
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,4,balanced,0.06826879978179931
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,4,balanced,0.07166399881243705
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,4,balanced,0.05198079980909824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,4,balanced,0.04264959953725338
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,4,balanced,0.07032320126891137
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,4,balanced,0.05405759997665882
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,4,balanced,0.04158400036394596
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,4,balanced,0.06998400203883648
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,4,balanced,0.053011199831962584
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,4,balanced,0.04158079996705055
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,4,balanced,0.06740479953587056
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,4,balanced,0.06968960016965867
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,4,balanced,0.0507520005106926
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,4,balanced,0.04288639947772026
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,4,balanced,0.10089600011706353
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,4,balanced,0.054019199684262276
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,4,balanced,0.04051840007305145
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,4,balanced,0.06782400012016296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,4,balanced,0.05248000025749207
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,4,balanced,0.0457311999052763
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,4,balanced,0.06594559997320175
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,4,balanced,0.06961920037865639
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,4,balanced,0.05343360044062138
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,4,balanced,0.041756799817085265
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,4,balanced,0.06772160008549691
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,4,balanced,0.051519999653100966
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,4,balanced,0.0409280002117157
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,4,balanced,0.06892160102725028
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,4,balanced,0.05244800001382828
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,4,balanced,0.043830399960279466
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,4,balanced,0.06786560080945492
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,4,balanced,0.07286399900913239
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,2,balanced,0.06153919957578182
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,2,balanced,0.052207999676465985
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,2,balanced,0.08140160068869591
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,2,balanced,0.06307840012013913
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,2,balanced,0.05157120004296303
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,2,balanced,0.08552640005946159
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.062323199957609175
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.05824639908969402
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.08336959928274154
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.08609279990196228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,2,balanced,0.0639552004635334
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,2,balanced,0.0481759998947382
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,2,balanced,0.0813216008245945
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,2,balanced,0.06575680002570153
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,2,balanced,0.0498816005885601
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,2,balanced,0.08256960064172744
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.06149440109729767
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.053308799490332605
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.08159679919481277
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.08062399998307228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,4,balanced,0.05343999974429607
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,4,balanced,0.043929599970579145
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,4,balanced,0.07001919895410538
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,2,balanced,0.06254079975187779
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,2,balanced,0.05066240057349205
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,2,balanced,0.0807903990149498
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,2,balanced,0.06153920069336891
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,2,balanced,0.051737600937485695
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,2,balanced,0.08207680061459541
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.06398719996213913
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.0526528000831604
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.08163520023226738
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.08081600069999695
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,2,balanced,0.060496000200510026
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,2,balanced,0.047971200197935104
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,2,balanced,0.07701439931988716
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,2,balanced,0.07805760055780411
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,2,balanced,0.06660799980163574
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,2,balanced,0.07569600045680999
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.07532479912042618
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.06415360048413277
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.07494400069117546
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.07612159997224807
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,2,balanced,0.07697600051760674
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,2,balanced,0.06322560086846352
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,2,balanced,0.07634559944272042
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,4,balanced,0.051452799886465075
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,2,balanced,0.07214719988405704
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,4,balanced,0.04140800014138222
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,2,balanced,0.06244160085916519
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,4,balanced,0.06794879995286465
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,2,balanced,0.07215680107474327
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.07519680000841618
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.05971520058810711
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.07627200037240982
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.0752191998064518
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,2,balanced,0.07023359984159469
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,2,balanced,0.06378240063786507
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,2,balanced,0.09792320057749748
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,2,balanced,0.07425599917769432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,2,balanced,0.05566399991512298
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,2,balanced,0.09125119969248771
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.06787839867174625
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.055212799832224844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.08536960184574127
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.0872064009308815
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,2,balanced,0.06538880057632923
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,2,balanced,0.05522879958152771
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,2,balanced,0.08650879934430122
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,4,balanced,0.05180159956216812
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,4,balanced,0.0443039994686842
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,4,balanced,0.06845439970493317
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,4,balanced,0.07168960049748421
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,2,balanced,0.06258879899978638
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,2,balanced,0.05368959978222847
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,2,balanced,0.08717760071158409
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.06124800033867359
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.05665600001811981
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.08684799969196319
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.0849600002169609
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,2,balanced,0.060451199486851695
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,2,balanced,0.05013439953327179
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,2,balanced,0.08664320036768913
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,2,balanced,0.059494400024414064
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,2,balanced,0.05213439986109734
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,2,balanced,0.08476480022072792
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.05735039971768856
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.052534399926662444
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.08242239952087402
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.08006399944424629
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,4,balanced,0.05019519962370396
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,4,balanced,0.048838399723172186
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,4,balanced,0.07309760004281998
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,2,balanced,0.05972799956798554
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,2,balanced,0.05176640003919601
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,2,balanced,0.08060159906744957
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,2,balanced,0.061686400324106216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,2,balanced,0.04787200018763542
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,2,balanced,0.07973119989037514
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.06108479984104633
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.051948800683021545
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.08132160007953644
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.0828224003314972
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,2,balanced,0.06102720014750958
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,2,balanced,0.047795199975371364
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,2,balanced,0.07783040031790733
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,2,balanced,0.057625599578022955
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,2,balanced,0.04767039977014065
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,2,balanced,0.0807455986738205
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,4,balanced,0.05437759980559349
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,4,balanced,0.04392320066690445
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,4,balanced,0.07060799971222878
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.06033920012414455
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.05135039947926998
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.08221120089292526
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.07679680064320564
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,2,balanced,0.06224000044167042
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,2,balanced,0.047887999564409256
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,2,balanced,0.08129279986023903
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,2,balanced,0.06507520079612732
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,2,balanced,0.047993600368499756
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,2,balanced,0.07897280007600785
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,4,balanced,0.05588479973375797
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,4,balanced,0.04289920032024384
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,4,balanced,0.06968640014529229
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,4,balanced,0.07751679942011833
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.05738880038261414
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.04808639995753765
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.07654720097780228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.07903039902448654
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,2,balanced,0.054787199944257736
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,2,balanced,0.04584320038557053
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,2,balanced,0.0809919998049736
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,2,balanced,0.05773759931325913
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,2,balanced,0.04592959992587566
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,2,balanced,0.07791360020637512
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.05849279947578907
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.04912639930844307
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.07895359843969345
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.08372480124235153
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,4,balanced,0.050710399821400645
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,4,balanced,0.05019199997186661
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,4,balanced,0.08207359910011292
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,2,balanced,0.05920960046350956
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,2,balanced,0.0493440005928278
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,2,balanced,0.08548479974269867
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,2,balanced,0.05767360106110573
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,2,balanced,0.04702400006353855
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,2,balanced,0.07683520093560219
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.05447999984025955
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.04997759982943535
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.08003840148448944
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.07863679900765419
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,2,balanced,0.06290880031883717
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,2,balanced,0.05217920020222664
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,2,balanced,0.07581119984388351
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,4,balanced,0.05681280046701431
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,4,balanced,0.042614400014281274
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,4,balanced,0.08032319992780686
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,2,balanced,0.06185600012540817
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,2,balanced,0.04697919972240925
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,2,balanced,0.08022399917244911
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.05544639937579632
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.05113920047879219
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.07884160056710243
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.07679360061883926
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,4,balanced,0.052614400163292885
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,4,balanced,0.04664320014417171
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,4,balanced,0.07981440052390099
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,4,balanced,0.09374400079250336
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,2,balanced,0.054076800122857094
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,2,balanced,0.05141120031476021
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,2,balanced,0.08339519947767257
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,2,balanced,0.055564799904823305
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,2,balanced,0.04991360083222389
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,2,balanced,0.07807359918951988
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.05655680038034916
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.04947200007736683
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.07880639955401421
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.08233280032873154
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,4,balanced,0.052825600281357764
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,4,balanced,0.05531839989125729
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,4,balanced,0.09348160102963447
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,2,balanced,0.05840959995985031
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,2,balanced,0.0504320003092289
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,2,balanced,0.07956480011343955
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,2,balanced,0.055039999261498454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,2,balanced,0.04805759973824024
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,2,balanced,0.07674239948391914
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.055747199803590775
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.049715199321508405
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.0806719996035099
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.08433919996023179
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,4,balanced,0.05204160064458847
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,4,balanced,0.04535040073096752
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,4,balanced,0.09345920011401176
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,2,balanced,0.05345280058681965
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,2,balanced,0.04975039996206761
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,2,balanced,0.08145599961280822
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,2,balanced,0.05767359957098961
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,2,balanced,0.04879679977893829
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,2,balanced,0.07780800014734268
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,4,balanced,0.04915199987590313
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,4,balanced,0.046063999831676486
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,4,balanced,0.09292159974575043
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,4,balanced,0.10878720059990883
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.06115839928388596
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.05266559943556785
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.0791423998773098
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.09379200041294097
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,2,balanced,0.056387200206518176
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,2,balanced,0.057148800045251844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,2,balanced,0.09387520030140876
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,4,balanced,0.0541056003421545
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,4,balanced,0.06667840033769608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,4,balanced,0.11856320053339005
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,2,balanced,0.061216000095009805
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,2,balanced,0.051468800753355026
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,2,balanced,0.09183359891176224
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.05903679989278317
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.05428159981966019
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.09246079921722412
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.11706559956073762
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,4,balanced,0.051423999667167666
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,4,balanced,0.049452800303697586
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,4,balanced,0.11917439997196197
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,2,balanced,0.05700479969382286
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,2,balanced,0.06468800120055676
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,2,balanced,0.10843840017914771
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,2,balanced,0.058707200735807416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,2,balanced,0.057308799400925634
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,2,balanced,0.10803839936852455
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,4,balanced,0.05422079972922802
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,4,balanced,0.05210560001432896
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,4,balanced,0.1183071993291378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,4,balanced,0.13779199868440628
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.05626560002565384
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.060358399897813796
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.10605759993195533
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.13967999964952468
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,4,balanced,0.05392640009522438
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,4,balanced,0.07561280131340027
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,4,balanced,0.14475519806146622
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,2,balanced,0.059052799642086026
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,2,balanced,0.07764159962534904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,2,balanced,0.13795200139284133
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,2,balanced,0.059756799787282947
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,2,balanced,0.06601599901914597
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,2,balanced,0.13627520054578782
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,4,balanced,0.05386560037732124
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,4,balanced,0.054896000400185586
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,4,balanced,0.14344960004091262
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.055711999908089636
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.06936319917440414
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.1391168013215065
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.18199680149555206
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,4,balanced,0.05462080053985119
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,4,balanced,0.05744959972798824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,4,balanced,0.14280000030994416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,4,balanced,0.16518079936504365
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,2,balanced,0.06064959987998009
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,2,balanced,0.0907744012773037
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,2,balanced,0.16547520011663436
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,16,balanced,0.06856639944016933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,16,balanced,0.06015999987721443
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,16,balanced,0.06874239966273307
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,16,balanced,0.07111360020935535
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,16,balanced,0.05457280017435551
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,16,balanced,0.06711040064692497
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,16,balanced,0.06873599961400031
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,16,balanced,0.056307199597358706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,16,balanced,0.07454079911112785
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,16,balanced,0.08077439963817597
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,16,balanced,0.06939519979059697
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,16,balanced,0.05418879985809326
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,16,balanced,0.07648960053920746
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,16,balanced,0.06658240035176277
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,16,balanced,0.05740160010755062
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,16,balanced,0.07514560073614121
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,16,balanced,0.06538240052759647
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,16,balanced,0.05507519990205765
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,16,balanced,0.06863999962806702
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,16,balanced,0.0762367993593216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,16,balanced,0.06380799971520901
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,16,balanced,0.05762560032308102
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,16,balanced,0.0744928002357483
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,16,balanced,0.06134080104529858
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,16,balanced,0.04888640008866787
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,16,balanced,0.06860800087451935
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,16,balanced,0.06555840075016021
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,16,balanced,0.05255359970033169
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,16,balanced,0.07136959917843341
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,16,balanced,0.07078079916536809
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,2,balanced,0.06034559980034828
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,2,balanced,0.07302400022745133
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,2,balanced,0.16636160016059875
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,16,balanced,0.06524480022490024
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,16,balanced,0.05034560039639473
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,16,balanced,0.07628160044550895
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,16,balanced,0.06538239978253842
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,16,balanced,0.05417600013315678
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,16,balanced,0.0780288003385067
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,16,balanced,0.06180800013244152
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,16,balanced,0.05244160033762455
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,16,balanced,0.0786495991051197
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,16,balanced,0.07748479992151261
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,16,balanced,0.06153280176222324
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,16,balanced,0.05321920067071915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,16,balanced,0.07727040126919746
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,16,balanced,0.06108799986541271
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,16,balanced,0.05084160007536411
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,16,balanced,0.07742720022797585
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,16,balanced,0.0630560003221035
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,16,balanced,0.05191999971866608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,16,balanced,0.07657279968261718
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,16,balanced,0.076265599578619
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,4,balanced,0.060895999521017076
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,4,balanced,0.0953855998814106
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,4,balanced,0.19626559913158417
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,16,balanced,0.06279039978981019
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,16,balanced,0.051465599611401555
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,16,balanced,0.08035839945077897
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,16,balanced,0.06085759997367859
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,16,balanced,0.053094399720430376
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,16,balanced,0.07895679920911788
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,16,balanced,0.06055359989404678
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,16,balanced,0.05404479950666428
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,16,balanced,0.07956480011343955
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,16,balanced,0.07847040072083473
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,16,balanced,0.06391040012240409
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,16,balanced,0.052937598899006844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,16,balanced,0.07664960101246834
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,16,balanced,0.06249920055270195
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,16,balanced,0.05232320055365562
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,16,balanced,0.07643520012497902
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,16,balanced,0.05908800028264523
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,16,balanced,0.05171839892864227
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,16,balanced,0.07744319960474969
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,16,balanced,0.07531520053744316
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,16,balanced,0.06043199934065342
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,16,balanced,0.05461440086364746
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,16,balanced,0.09068480059504509
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,16,balanced,0.0622783999890089
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,16,balanced,0.0540800005197525
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,16,balanced,0.07627200111746787
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,16,balanced,0.060207999870181085
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,16,balanced,0.053011199831962584
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,16,balanced,0.07671360149979592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,16,balanced,0.07925119921565056
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,16,balanced,0.0599263995885849
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,16,balanced,0.04938879981637001
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,16,balanced,0.07698559984564782
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,16,balanced,0.0586303997784853
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,16,balanced,0.04999040067195892
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,16,balanced,0.07670719921588898
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,16,balanced,0.06251199953258038
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,16,balanced,0.05174399949610233
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,16,balanced,0.07636479958891869
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,16,balanced,0.07574720084667205
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,16,balanced,0.05898559950292111
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,16,balanced,0.05143040008842945
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,16,balanced,0.07651840075850487
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,16,balanced,0.0606623999774456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,16,balanced,0.0503040000796318
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,16,balanced,0.07664000019431114
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.058105599507689475
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.07823040187358857
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.16399359852075576
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.22299199998378755
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,16,balanced,0.061753600090742114
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,16,balanced,0.05407680049538612
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,16,balanced,0.0763935998082161
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,16,balanced,0.07743999883532524
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,16,balanced,0.06028799973428249
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,16,balanced,0.07052160017192363
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,16,balanced,0.09076160043478013
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,16,balanced,0.0650624006986618
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,16,balanced,0.053920000046491626
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,16,balanced,0.08697599992156028
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,16,balanced,0.0611904002726078
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,16,balanced,0.05825600065290928
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,16,balanced,0.08643520027399063
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,16,balanced,0.08687360063195229
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,16,balanced,0.06844160035252571
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,16,balanced,0.05846079997718334
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,16,balanced,0.08017919957637787
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,16,balanced,0.0582847997546196
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,16,balanced,0.048739200085401536
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,16,balanced,0.07724159955978394
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,16,balanced,0.06360319927334786
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,16,balanced,0.05420159921050072
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,16,balanced,0.07921279966831207
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,16,balanced,0.07712960094213486
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,4,balanced,0.05911360010504722
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,16,balanced,0.06172480061650276
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,4,balanced,0.06555840075016021
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,16,balanced,0.05652800016105175
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,4,balanced,0.19678079932928086
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,16,balanced,0.07575039938092232
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,16,balanced,0.0582847997546196
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,16,balanced,0.04949120059609413
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,16,balanced,0.07682240009307861
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,16,balanced,0.059036800265312196
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,16,balanced,0.05425599925220013
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,16,balanced,0.08040639981627465
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,16,balanced,0.07494079992175103
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,16,balanced,0.06114240065217018
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,16,balanced,0.059046399965882304
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,16,balanced,0.07905920073390008
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,16,balanced,0.061430399492383006
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,16,balanced,0.05012800060212612
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,16,balanced,0.0819263994693756
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,2,balanced,0.0629343993961811
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,2,balanced,0.12344640046358109
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,2,balanced,0.22431039959192275
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,16,balanced,0.06460799984633922
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,16,balanced,0.055136000365018846
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,16,balanced,0.08172799944877625
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,16,balanced,0.08038079962134362
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,16,balanced,0.06260160095989704
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,16,balanced,0.057939200475811956
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,16,balanced,0.07816639989614486
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,16,balanced,0.0629184003919363
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,16,balanced,0.05146880000829697
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,16,balanced,0.08238400220870971
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,16,balanced,0.0624159999191761
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,16,balanced,0.05295999981462955
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,16,balanced,0.08273279964923859
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,16,balanced,0.08211840018630027
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,16,balanced,0.059510400891304015
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,16,balanced,0.05727359987795353
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,16,balanced,0.08252799957990646
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,4,balanced,0.05887680016458034
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,4,balanced,0.06791360005736351
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,4,balanced,0.19447359889745713
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,4,balanced,0.22468159794807435
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,16,balanced,0.059043199941515925
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,16,balanced,0.05247360020875931
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,16,balanced,0.07951679974794387
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,16,balanced,0.06050879918038845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,16,balanced,0.052937600389122964
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,16,balanced,0.08337280005216599
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,16,balanced,0.08089600130915642
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,2,balanced,0.06455360054969787
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,2,balanced,0.09177599996328353
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,2,balanced,0.22102719992399217
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,16,balanced,0.06543039940297604
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,16,balanced,0.06071040071547031
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,16,balanced,0.08803200051188469
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,16,balanced,0.06222720108926296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,16,balanced,0.053856000676751135
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,16,balanced,0.08453120067715644
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,16,balanced,0.06460800021886826
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,16,balanced,0.05485760048031807
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,16,balanced,0.08535040020942689
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,16,balanced,0.08491840064525605
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,16,balanced,0.06777919940650463
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,16,balanced,0.0647071998566389
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,16,balanced,0.0959136001765728
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.0627423994243145
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.1003584012389183
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.22189119905233384
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.3114560008049011
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,16,balanced,0.065718400105834
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,16,balanced,0.05552320033311844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,16,balanced,0.09908800050616265
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,4,balanced,0.06718079932034016
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,4,balanced,0.11467519998550416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,4,balanced,0.24538239985704421
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,16,balanced,0.0658912006765604
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,16,balanced,0.0562400009483099
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,16,balanced,0.09422399997711181
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,16,balanced,0.09564799964427947
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,16,balanced,0.06436800062656403
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,16,balanced,0.07434239983558655
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,16,balanced,0.11024959981441498
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,16,balanced,0.06612479947507381
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,16,balanced,0.058828800544142724
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,16,balanced,0.10856639966368675
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,2,balanced,0.07032959945499898
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,2,balanced,0.1492576003074646
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,2,balanced,0.2827392041683197
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,16,balanced,0.06460480056703091
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,16,balanced,0.05804159976541996
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,16,balanced,0.10843200013041496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,16,balanced,0.10578560009598732
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,4,balanced,0.06702079996466637
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,4,balanced,0.08007039874792099
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,4,balanced,0.24453120231628417
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,16,balanced,0.07019519992172718
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,16,balanced,0.08334719985723496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,16,balanced,0.13686719983816148
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,8,balanced,0.06933120042085647
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,8,balanced,0.054096000641584395
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,8,balanced,0.07768639996647835
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,8,balanced,0.062060799822211264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,8,balanced,0.05302719995379448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,8,balanced,0.0770079992711544
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,32,balanced,0.07469759956002235
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,32,balanced,0.06871359944343566
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,8,balanced,0.0634527999907732
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,32,balanced,0.07837119996547699
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,8,balanced,0.05378240011632442
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,8,balanced,0.07596159800887108
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,8,balanced,0.07965760007500648
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,8,balanced,0.06348160058259963
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,8,balanced,0.051744000241160396
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,8,balanced,0.07578559964895248
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,32,balanced,0.07135679945349693
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,32,balanced,0.061366400122642516
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,8,balanced,0.059996800124645235
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,32,balanced,0.07942080050706864
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,8,balanced,0.05044800043106079
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,8,balanced,0.07390720024704933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,8,balanced,0.05883840024471283
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,8,balanced,0.0513408001512289
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,8,balanced,0.07790080010890961
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,8,balanced,0.07597440183162689
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,32,balanced,0.07237439900636673
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,32,balanced,0.05751039907336235
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,32,balanced,0.08007360026240348
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,32,balanced,0.07876479998230934
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,8,balanced,0.05611520037055016
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,8,balanced,0.05329280011355877
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,8,balanced,0.0722079999744892
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,32,balanced,0.06756480038166046
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,32,balanced,0.054764799401164056
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,8,balanced,0.06616960018873215
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,32,balanced,0.07959680035710334
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,8,balanced,0.048982400074601175
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,16,balanced,0.06415359973907471
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,8,balanced,0.07513600066304207
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,16,balanced,0.06200640015304089
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,16,balanced,0.13367680162191392
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,32,balanced,0.06893440037965774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,32,balanced,0.05654399916529655
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,32,balanced,0.07509120032191277
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,8,balanced,0.05876480005681515
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,8,balanced,0.048748799785971644
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,8,balanced,0.07571200057864189
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,8,balanced,0.07614079937338829
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,32,balanced,0.06601919941604137
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,32,balanced,0.05464640036225319
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,32,balanced,0.07708479985594749
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,32,balanced,0.07685439959168434
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,8,balanced,0.05880319960415363
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,8,balanced,0.04827200062572956
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,8,balanced,0.08129920065402985
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,32,balanced,0.06595200039446354
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,32,balanced,0.05313920043408871
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,32,balanced,0.07846400141716003
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,8,balanced,0.05861120000481605
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,8,balanced,0.04927679896354675
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,8,balanced,0.0765087991952896
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,2,balanced,0.07026880010962486
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,32,balanced,0.0671135999262333
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,2,balanced,0.12066880017518997
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,32,balanced,0.055087999626994136
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,2,balanced,0.28196479976177213
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,32,balanced,0.07373120039701461
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,8,balanced,0.05838399939239025
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,8,balanced,0.052803199738264084
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,8,balanced,0.0733984000980854
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,8,balanced,0.08099199905991554
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,32,balanced,0.06955840066075325
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,32,balanced,0.0532768003642559
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,32,balanced,0.07799680009484292
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,32,balanced,0.08111360073089599
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,8,balanced,0.05775039978325367
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,8,balanced,0.04946880005300045
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,8,balanced,0.07918400019407272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,32,balanced,0.06541119888424873
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,32,balanced,0.05268159955739975
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,32,balanced,0.07588160037994385
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,8,balanced,0.056668800115585324
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,8,balanced,0.04700479954481125
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,8,balanced,0.07567679956555366
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,32,balanced,0.06833279989659787
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,32,balanced,0.04999679960310459
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,32,balanced,0.07669440060853958
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,8,balanced,0.058937599509954454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,8,balanced,0.048294400423765184
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,8,balanced,0.07455679848790168
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,8,balanced,0.07235520109534263
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,32,balanced,0.0631584007292986
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,32,balanced,0.0524447999894619
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,32,balanced,0.08060480058193206
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,32,balanced,0.07951999977231025
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,8,balanced,0.055183999612927435
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,8,balanced,0.046953600272536276
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,8,balanced,0.07228479981422424
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,32,balanced,0.0678751990199089
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,32,balanced,0.053360000252723694
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,32,balanced,0.08240320011973382
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,8,balanced,0.058243200182914734
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,8,balanced,0.049139199405908586
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,8,balanced,0.07469440028071403
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,32,balanced,0.06730240024626255
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,32,balanced,0.05341439954936504
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,32,balanced,0.07921599969267845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,8,balanced,0.05772479958832264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,8,balanced,0.04808639995753765
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,8,balanced,0.07326719984412193
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,32,balanced,0.061833600699901584
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,8,balanced,0.0745471991598606
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,32,balanced,0.05495679974555969
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,32,balanced,0.0759583979845047
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,32,balanced,0.0786976009607315
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,8,balanced,0.057145599648356436
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,32,balanced,0.061711999028921126
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,8,balanced,0.045177600160241124
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,32,balanced,0.06013120003044605
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,8,balanced,0.07374719977378845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,32,balanced,0.08368959948420525
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,32,balanced,0.06431680023670197
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,32,balanced,0.054447999224066734
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,32,balanced,0.07833599969744683
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,8,balanced,0.05765440016984939
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,8,balanced,0.04799679964780808
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,8,balanced,0.0711039997637272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,4,balanced,0.06613759957253933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,4,balanced,0.07600959986448289
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,4,balanced,0.24490240067243577
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,32,balanced,0.0637184001505375
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,4,balanced,0.28074879944324493
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,32,balanced,0.05212159976363182
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,8,balanced,0.056044799834489824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,32,balanced,0.07928960099816322
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,8,balanced,0.04903679974377155
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,32,balanced,0.07902720049023629
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,8,balanced,0.0715551994740963
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,8,balanced,0.07035520076751708
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,32,balanced,0.06324159987270832
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,32,balanced,0.0606368001550436
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,8,balanced,0.05710720047354698
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,32,balanced,0.08102400079369546
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,8,balanced,0.04614719934761524
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,8,balanced,0.07485760003328323
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,32,balanced,0.05910079926252365
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,32,balanced,0.05479039959609509
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,32,balanced,0.07852160036563874
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,8,balanced,0.0562720000743866
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,8,balanced,0.046512000262737274
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,8,balanced,0.07376639991998672
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,32,balanced,0.061171199381351474
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,32,balanced,0.053036799654364586
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,32,balanced,0.07748799994587899
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,32,balanced,0.08035200014710427
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,8,balanced,0.05881600081920624
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,8,balanced,0.04686079993844032
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,8,balanced,0.07091839984059334
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,8,balanced,0.07293759882450104
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,32,balanced,0.06172800026834011
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,32,balanced,0.05872000083327293
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,32,balanced,0.07788159996271134
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,8,balanced,0.054022399708628654
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,16,balanced,0.0672800000756979
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,8,balanced,0.04504639990627766
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,16,balanced,0.058316799625754354
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,8,balanced,0.07182719931006432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,16,balanced,0.13537279963493348
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,16,balanced,0.1320127986371517
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,32,balanced,0.06282560005784035
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,32,balanced,0.05261440053582191
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,32,balanced,0.07998720034956933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,8,balanced,0.055894399806857106
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,8,balanced,0.044604799896478656
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,8,balanced,0.07021439895033836
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,32,balanced,0.06203840039670468
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,32,balanced,0.055196800455451014
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,32,balanced,0.08024960085749626
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,32,balanced,0.0819072000682354
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,8,balanced,0.05531199984252453
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,8,balanced,0.048662399500608446
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,8,balanced,0.0726432003080845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,8,balanced,0.0738816000521183
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,32,balanced,0.0619488000869751
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,32,balanced,0.06129599958658218
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,32,balanced,0.08085760027170182
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,8,balanced,0.053564800694584844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,8,balanced,0.04560639970004558
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,8,balanced,0.073369599878788
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,32,balanced,0.06323839984834194
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,32,balanced,0.05335040017962456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,32,balanced,0.07928960025310516
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,8,balanced,0.058278399705886844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,8,balanced,0.046108799800276754
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,8,balanced,0.07145920023322105
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,32,balanced,0.06021120026707649
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,32,balanced,0.05671039968729019
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,32,balanced,0.07767999991774559
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,32,balanced,0.08056320026516914
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,8,balanced,0.05584320016205311
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,8,balanced,0.048476799950003625
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,8,balanced,0.07383360043168068
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,8,balanced,0.07909119948744774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,32,balanced,0.06377599947154522
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,32,balanced,0.05702400058507919
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,32,balanced,0.07550400123000145
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,8,balanced,0.05656959936022758
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,8,balanced,0.04894400015473366
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,32,balanced,0.06138239949941635
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,8,balanced,0.07348480075597763
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,32,balanced,0.05304960012435913
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,32,balanced,0.08071679770946502
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,32,balanced,0.0583456002175808
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,32,balanced,0.05302079990506172
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,32,balanced,0.07963840141892434
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,8,balanced,0.05569280013442039
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,32,balanced,0.07963839992880821
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,8,balanced,0.04679680019617081
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,8,balanced,0.0732000008225441
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,32,balanced,0.061788799986243245
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,8,balanced,0.055596799775958064
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,32,balanced,0.05978239923715591
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,8,balanced,0.052252800390124324
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,32,balanced,0.07811839953064918
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,8,balanced,0.07126400023698806
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,8,balanced,0.07572799921035767
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,32,balanced,0.060912000015378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,32,balanced,0.05078400000929832
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,32,balanced,0.09880320057272911
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,8,balanced,0.056460800021886824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,8,balanced,0.04730559960007667
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,8,balanced,0.07826559990644455
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,32,balanced,0.05754239968955517
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,32,balanced,0.05352640002965927
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,32,balanced,0.08030719980597496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,32,balanced,0.07753600031137467
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,8,balanced,0.05772479996085167
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,8,balanced,0.04690559916198254
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,8,balanced,0.07206400111317635
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,32,balanced,0.06198719963431358
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,32,balanced,0.050067199394106865
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,32,balanced,0.07362879961729049
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,8,balanced,0.05694080069661141
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,8,balanced,0.04679039977490902
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,8,balanced,0.07485440000891685
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,8,balanced,0.07332159876823426
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,16,balanced,0.06857599951326847
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,16,balanced,0.09418239966034889
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,16,balanced,0.1621471971273422
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,32,balanced,0.05568960011005401
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,32,balanced,0.046428800001740454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,32,balanced,0.07599680125713348
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,8,balanced,0.054249600321054456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,8,balanced,0.047728000208735466
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,8,balanced,0.0732863999903202
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,32,balanced,0.05784000046551228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,32,balanced,0.05103679932653904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,32,balanced,0.07550400048494339
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,32,balanced,0.07666239961981773
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.07000000029802322
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.12111999839544296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.2791648000478745
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.3963712006807327
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,8,balanced,0.055587200075387956
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,8,balanced,0.046912000700831416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,8,balanced,0.07267519980669021
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,32,balanced,0.05926719978451729
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,32,balanced,0.053119999915361406
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,32,balanced,0.07981760054826736
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,8,balanced,0.055936000868678094
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,8,balanced,0.04899520017206669
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,8,balanced,0.0756000004708767
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,8,balanced,0.07568320035934448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,32,balanced,0.06042879968881607
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,32,balanced,0.05028799995779991
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,32,balanced,0.07564800083637238
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,8,balanced,0.05339200012385845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,8,balanced,0.04885439984500408
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,8,balanced,0.07792320027947426
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,32,balanced,0.05806399919092655
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,32,balanced,0.05114879943430424
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,32,balanced,0.07421760037541389
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,32,balanced,0.07422720044851303
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,8,balanced,0.05499200001358986
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,8,balanced,0.046352000534534456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,8,balanced,0.07255999967455865
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,32,balanced,0.0565311998128891
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,32,balanced,0.055775999650359155
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,32,balanced,0.07777919992804527
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,8,balanced,0.05332159996032715
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,8,balanced,0.048470400646328925
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,8,balanced,0.07043519914150238
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,8,balanced,0.07300800085067749
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,32,balanced,0.059110399708151816
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,32,balanced,0.05332160033285618
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,32,balanced,0.08106560111045838
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,8,balanced,0.0576191995292902
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,8,balanced,0.04909119941294193
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,8,balanced,0.07619199976325035
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,32,balanced,0.05818880014121532
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,32,balanced,0.05058559961616993
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,32,balanced,0.07663360014557838
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,32,balanced,0.07844479978084565
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,16,balanced,0.06982399895787239
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,16,balanced,0.0710271991789341
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,16,balanced,0.16048320084810258
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,8,balanced,0.05447360016405582
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,8,balanced,0.05299839936196804
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,8,balanced,0.07556799873709678
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,32,balanced,0.060239999741315844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,32,balanced,0.060662399232387546
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,32,balanced,0.07411200031638146
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,8,balanced,0.055935999751091
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,8,balanced,0.04723200015723705
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,8,balanced,0.07286079972982407
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,8,balanced,0.07479679957032204
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,32,balanced,0.07398720048367977
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,32,balanced,0.05271680019795895
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,32,balanced,0.07692480012774468
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,8,balanced,0.05927999950945377
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,8,balanced,0.05
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,8,balanced,0.07493120059370995
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,32,balanced,0.061372799426317216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,32,balanced,0.05183360017836094
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,32,balanced,0.07613120004534721
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,32,balanced,0.07752320021390915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,4,balanced,0.07959679961204529
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,4,balanced,0.1583840012550354
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,4,balanced,0.344336000084877
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,8,balanced,0.06165119968354702
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,8,balanced,0.04703679978847504
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,8,balanced,0.07144959941506386
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,32,balanced,0.058355200290679934
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,32,balanced,0.06933440081775188
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,32,balanced,0.07792639955878258
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,8,balanced,0.05841599963605404
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,8,balanced,0.04750399999320507
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,8,balanced,0.07625280022621155
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,8,balanced,0.07520640045404434
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,32,balanced,0.060358400270342824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,32,balanced,0.052428800240159036
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,32,balanced,0.07840640023350716
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,16,balanced,0.06760959997773171
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,16,balanced,0.062211199477314946
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,16,balanced,0.1608831986784935
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,16,balanced,0.15434879958629608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,32,balanced,0.0604480005800724
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,32,balanced,0.05005120001733303
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,32,balanced,0.07734080031514168
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,32,balanced,0.07779519930481911
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,8,balanced,0.05549120008945465
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,8,balanced,0.05422400049865246
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,8,balanced,0.07804159894585609
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,8,balanced,0.0563424002379179
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,8,balanced,0.04635199978947639
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,8,balanced,0.07767680063843727
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,32,balanced,0.06315840035676956
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,32,balanced,0.0668576005846262
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,32,balanced,0.08145599961280822
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,8,balanced,0.06117760054767132
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,8,balanced,0.048204800114035606
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,8,balanced,0.07450559958815575
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,8,balanced,0.07779839932918549
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,32,balanced,0.06025599986314774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,32,balanced,0.05225599892437458
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,32,balanced,0.08149759918451309
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,2,balanced,0.08443520069122315
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,2,balanced,0.2065088003873825
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,2,balanced,0.3953503996133804
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,32,balanced,0.05879999957978725
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,32,balanced,0.04909119978547096
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,32,balanced,0.08022400066256523
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,32,balanced,0.07999999970197677
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,8,balanced,0.05507519952952862
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,8,balanced,0.05880319960415363
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,8,balanced,0.08920319899916648
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,32,balanced,0.0626144003123045
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,32,balanced,0.07197440080344678
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,32,balanced,0.0932639978826046
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,16,balanced,0.07285120040178299
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,8,balanced,0.05626239962875843
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,16,balanced,0.11919680014252662
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,8,balanced,0.05011199973523617
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,16,balanced,0.21706240028142929
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,8,balanced,0.08566080033779144
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,32,balanced,0.05992320068180561
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,32,balanced,0.05756479986011982
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,32,balanced,0.09399039968848229
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,8,balanced,0.058051199465990064
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,8,balanced,0.04880640022456646
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,8,balanced,0.08797439932823181
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,8,balanced,0.08848640024662018
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,32,balanced,0.06065279953181744
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,32,balanced,0.054383999481797216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,32,balanced,0.09327680096030236
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,32,balanced,0.09231359958648681
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,8,balanced,0.06300799995660782
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,8,balanced,0.06562239937484264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,8,balanced,0.1015807993710041
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,4,balanced,0.08037439957261086
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,4,balanced,0.10908159911632538
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,4,balanced,0.3450848013162613
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,32,balanced,0.06883199959993362
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,32,balanced,0.07210239991545678
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,32,balanced,0.10811200141906738
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,8,balanced,0.054422400146722796
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,8,balanced,0.05135359987616539
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,8,balanced,0.10117759928107262
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,16,balanced,0.07546559944748879
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,16,balanced,0.08248319923877716
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,16,balanced,0.21447360068559645
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,32,balanced,0.06266879960894585
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,32,balanced,0.056726400181651114
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,32,balanced,0.10273599922657013
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,8,balanced,0.05765760056674481
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,8,balanced,0.048595200106501577
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,8,balanced,0.10056639909744262
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,8,balanced,0.10144639983773232
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,2,balanced,0.08114560022950172
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,2,balanced,0.16994240134954453
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,2,balanced,0.39805439710617063
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,32,balanced,0.06276480071246623
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,32,balanced,0.05843200013041496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,32,balanced,0.10801919996738434
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,32,balanced,0.10796480029821395
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,8,balanced,0.059673600271344183
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,8,balanced,0.07388480007648468
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,8,balanced,0.12671039924025534
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,16,balanced,0.0787167988717556
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,16,balanced,0.07106559947133065
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,16,balanced,0.21414400339126588
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,16,balanced,0.20370559692382811
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,32,balanced,0.06660160087049008
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,32,balanced,0.08821759968996049
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,32,balanced,0.13497280031442643
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,8,balanced,0.06117120087146759
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,8,balanced,0.05817919969558716
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,8,balanced,0.12520320117473602
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,32,balanced,0.06411520019173622
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,32,balanced,0.06309440061450004
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,32,balanced,0.13364160060882568
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,4,balanced,0.0795008011162281
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,4,balanced,0.09693120047450066
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,4,balanced,0.34300160109996797
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,4,balanced,0.40140160024166105
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,8,balanced,0.056934400275349614
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,8,balanced,0.052143999934196474
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,8,balanced,0.12565439864993094
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,8,balanced,0.12477119788527488
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,32,balanced,0.06754559986293315
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,32,balanced,0.05854080021381378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,32,balanced,0.1317120000720024
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,32,balanced,0.12932480052113532
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.08132480010390282
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.16081600189208983
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.3939839988946915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.5699903905391693
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,16,balanced,0.07955520004034042
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,16,balanced,0.14179519861936568
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,16,balanced,0.27045120000839235
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,8,balanced,0.05999359972774983
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,8,balanced,0.08395199924707412
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,8,balanced,0.15315519869327546
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,32,balanced,0.06960640028119087
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,32,balanced,0.0975455991923809
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,32,balanced,0.1586911976337433
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,8,balanced,0.05995839945971966
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,8,balanced,0.0626751996576786
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,8,balanced,0.1513248011469841
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,32,balanced,0.06817280054092408
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,32,balanced,0.07111039981245995
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,32,balanced,0.15950080156326293
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,16,balanced,0.08027840033173561
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,16,balanced,0.09414719939231872
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,16,balanced,0.26828799545764925
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,8,balanced,0.057356799766421315
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,8,balanced,0.05360959991812706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,8,balanced,0.1508448004722595
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,8,balanced,0.14711040109395981
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,32,balanced,0.0661600012332201
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,32,balanced,0.06436479985713958
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,32,balanced,0.16061120182275773
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,32,balanced,0.15382720082998275
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,4,balanced,0.09197760000824928
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,4,balanced,0.19889599978923797
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,4,balanced,0.44788160026073454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,2,balanced,0.09505600035190583
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,2,balanced,0.26276480257511137
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,2,balanced,0.5110624015331269
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,8,balanced,0.06479040049016475
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,8,balanced,0.10566399991512299
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,8,balanced,0.20582720041275024
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,16,balanced,0.07895679920911788
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,16,balanced,0.07690240070223808
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,16,balanced,0.2666719973087311
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,16,balanced,0.2516608014702797
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,32,balanced,0.07170559987425804
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,32,balanced,0.12244800105690956
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,32,balanced,0.22014400213956833
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,8,balanced,0.06385599933564663
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,8,balanced,0.07152639999985695
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,8,balanced,0.204959999024868
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,32,balanced,0.07258560135960579
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,32,balanced,0.08304319903254509
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,32,balanced,0.21608960181474685
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,8,balanced,0.06323199942708016
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,8,balanced,0.06310080029070378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,8,balanced,0.20322239995002747
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,8,balanced,0.19476799964904784
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,4,balanced,0.0930527999997139
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,4,balanced,0.12788160070776938
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,4,balanced,0.4438816040754318
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,2,balanced,0.09458559900522232
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,2,balanced,0.2134304016828537
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,2,balanced,0.5094143927097321
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,16,balanced,0.09261759966611863
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,16,balanced,0.19014399945735933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,16,balanced,0.37428160309791564
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,32,balanced,0.07033920064568519
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,32,balanced,0.07391039952635765
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,32,balanced,0.22099519968032838
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,32,balanced,0.20318719744682312
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,8,balanced,0.06958719938993455
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,8,balanced,0.12705919966101648
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,8,balanced,0.2564863979816437
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,32,balanced,0.07806399911642074
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,32,balanced,0.1479487970471382
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,32,balanced,0.2720544010400772
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,16,balanced,0.08904320076107979
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,16,balanced,0.12005119919776916
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,16,balanced,0.37831680178642274
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.09388159886002541
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.2040096029639244
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.509059202671051
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.740230405330658
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,8,balanced,0.07190399914979935
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,8,balanced,0.08451199904084206
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,8,balanced,0.25552319884300234
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,4,balanced,0.09091200083494186
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,4,balanced,0.11886719837784768
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,4,balanced,0.44630720317363737
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,4,balanced,0.5144736111164093
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,32,balanced,0.08217600062489509
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,32,balanced,0.09527680054306983
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,32,balanced,0.27340159714221957
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,8,balanced,0.06954880058765411
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,8,balanced,0.0702015995979309
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,8,balanced,0.25420480221509933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,8,balanced,0.24192319959402084
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,16,balanced,0.09168320000171662
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,16,balanced,0.09615360125899315
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,16,balanced,0.37509440183639525
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,16,balanced,0.3449856013059616
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,32,balanced,0.09788799956440926
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,32,balanced,0.07871680036187172
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,32,balanced,0.27008000016212463
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,32,balanced,0.25071039944887163
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,8,balanced,0.08428480103611946
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,8,balanced,0.1746976003050804
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,8,balanced,0.3602271944284439
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,2,balanced,0.12226879894733429
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,2,balanced,0.3716959983110428
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,2,balanced,0.7381920099258423
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,32,balanced,0.09026560112833977
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,32,balanced,0.19166399985551835
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,32,balanced,0.37716799676418306
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,4,balanced,0.1174464002251625
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,4,balanced,0.27510400116443634
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,4,balanced,0.6472991943359375
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,16,balanced,0.10343040004372597
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,16,balanced,0.23438080102205278
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,16,balanced,0.47899200320243834
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,8,balanced,0.08319040164351463
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,8,balanced,0.10845119878649712
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,8,balanced,0.3575679987668991
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,32,balanced,0.08786239996552467
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,32,balanced,0.12157760038971901
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,32,balanced,0.38264640271663664
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,16,balanced,0.10648640096187592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,16,balanced,0.14522879868745803
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,16,balanced,0.4792191982269287
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,2,balanced,0.12332800030708313
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,2,balanced,0.2986687958240509
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,2,balanced,0.7371936023235321
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,8,balanced,0.08247359916567802
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,8,balanced,0.08404159918427467
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,8,balanced,0.3568159997463226
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,8,balanced,0.33932160437107084
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,4,balanced,0.11593599990010262
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,4,balanced,0.17347519993782043
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,4,balanced,0.6458559930324554
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,32,balanced,0.08742399886250496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,32,balanced,0.09581120014190674
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,32,balanced,0.3782431960105896
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,32,balanced,0.3457119971513748
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,16,balanced,0.10577919930219651
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,16,balanced,0.10913600027561188
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,16,balanced,0.4800896018743515
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,16,balanced,0.43739200532436373
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,8,balanced,0.09377600029110908
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,8,balanced,0.21504640132188796
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,8,balanced,0.46049919426441194
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,32,balanced,0.10157120153307915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,32,balanced,0.2446240022778511
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,32,balanced,0.4926559954881668
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,0.12112639993429183
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,0.2887295991182327
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,0.7356960058212281
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,1.0844352006912232
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,4,balanced,0.11816319823265076
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,4,balanced,0.16172159910202027
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,4,balanced,0.646451199054718
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,4,balanced,0.7494208097457886
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,8,balanced,0.09501120001077652
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,8,balanced,0.13228799998760224
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,8,balanced,0.45697600245475767
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,32,balanced,0.10266880020499229
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,32,balanced,0.15078080147504808
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,32,balanced,0.4892000019550323
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,16,balanced,0.12653439939022065
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,16,balanced,0.3257760047912598
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,16,balanced,0.6914752066135407
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,8,balanced,0.09335360080003738
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,8,balanced,0.09399999976158142
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,8,balanced,0.4602400004863739
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,8,balanced,0.4282688021659851
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,2,balanced,0.14630400091409684
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,2,balanced,0.4855423986911774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,2,balanced,0.9703839957714081
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,32,balanced,0.10748480036854743
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,32,balanced,0.1134783998131752
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,32,balanced,0.495279997587204
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,32,balanced,0.443638402223587
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,4,balanced,0.14301760345697404
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,4,balanced,0.3635872036218643
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,4,balanced,0.8499519944190979
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,16,balanced,0.1291264019906521
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,16,balanced,0.19259200245141983
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,16,balanced,0.6914624035358429
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,8,balanced,0.11803839951753617
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,8,balanced,0.298294398188591
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,8,balanced,0.6661440014839173
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,32,balanced,0.12837439998984337
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,32,balanced,0.3355263978242874
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,32,balanced,0.6940448045730591
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,2,balanced,0.1506303995847702
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,2,balanced,0.38844160437583924
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,2,balanced,0.9651616036891937
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,16,balanced,0.12909120097756385
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,16,balanced,0.1432224005460739
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,16,balanced,0.6912576019763946
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,16,balanced,0.6291936039924622
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,4,balanced,0.1405312016606331
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,4,balanced,0.2208639994263649
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,4,balanced,0.8453792035579681
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,8,balanced,0.12146559953689576
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,8,balanced,0.17714240103960038
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,8,balanced,0.663641607761383
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,32,balanced,0.12121919989585876
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,32,balanced,0.19688960164785385
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,32,balanced,0.6943135917186737
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,0.1494688019156456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,0.370412802696228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,0.9623583972454071
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,1.4309567928314209
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,8,balanced,0.12006399929523467
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,8,balanced,0.11919359937310219
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,8,balanced,0.6645568013191223
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,8,balanced,0.6183520078659057
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,16,balanced,0.15143040120601653
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,16,balanced,0.4169663995504379
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,16,balanced,0.9032415926456452
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,4,balanced,0.13978559970855714
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,4,balanced,0.20281600058078766
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,4,balanced,0.8488256096839905
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,4,balanced,0.9798879981040954
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,32,balanced,0.12085760086774826
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,32,balanced,0.14395520240068435
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,32,balanced,0.6925407886505127
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,32,balanced,0.6260031998157501
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,2,balanced,0.17567359954118728
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,2,balanced,0.5950464010238647
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,2,balanced,1.2016608119010925
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,8,balanced,0.14567360132932664
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,8,balanced,0.3813759982585907
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,8,balanced,0.8704671919345855
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,16,balanced,0.1574528008699417
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,16,balanced,0.2418175995349884
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,16,balanced,0.9024991929531098
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,4,balanced,0.16594560146331788
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,4,balanced,0.43959360122680663
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,4,balanced,1.0563199996948243
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,32,balanced,0.14568639695644378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,32,balanced,0.43160640001296996
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,32,balanced,0.9085376024246216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,8,balanced,0.14253120124340057
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,8,balanced,0.22206720113754272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,8,balanced,0.869839996099472
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,2,balanced,0.1759872019290924
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,2,balanced,0.47416639924049375
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,2,balanced,1.2000831961631775
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,16,balanced,0.1563712015748024
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,16,balanced,0.16927359998226166
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,16,balanced,0.9060959994792939
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,16,balanced,0.808492797613144
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,32,balanced,0.14567679762840272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,32,balanced,0.2508799985051155
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,32,balanced,0.9093728065490723
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,4,balanced,0.16530559957027435
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,4,balanced,0.26664000153541567
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,4,balanced,1.0538527965545654
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,8,balanced,0.14413119852542877
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,8,balanced,0.14383040070533754
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,8,balanced,0.8712000131607056
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,8,balanced,0.798003202676773
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,0.1743711993098259
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,0.45260800123214723
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,1.1975552082061767
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,1.7779424071311951
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,32,balanced,0.1454943984746933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,32,balanced,0.17915200144052507
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,32,balanced,0.9070912003517151
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,32,balanced,0.8284319877624512
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,16,balanced,0.17845440059900283
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,16,balanced,0.5090816080570221
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,16,balanced,1.1180703997612
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,4,balanced,0.16491200029850006
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,4,balanced,0.2462400019168854
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,4,balanced,1.0563935875892638
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,4,balanced,1.222489583492279
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,8,balanced,0.16909440010786056
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,8,balanced,0.4680575966835022
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,8,balanced,1.080505621433258
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,32,balanced,0.17051199972629547
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,32,balanced,0.5267231941223145
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,32,balanced,1.1314048171043396
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,16,balanced,0.1739551991224289
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,16,balanced,0.28707840144634245
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,16,balanced,1.1206656098365784
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,2,balanced,0.2510591998696327
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,2,balanced,0.9225216031074523
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,2,balanced,1.9487743973731995
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,8,balanced,0.16704960018396378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,8,balanced,0.2706528067588806
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,8,balanced,1.079472005367279
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,4,balanced,0.24220480024814606
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,4,balanced,0.6870911955833435
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,4,balanced,1.6609600067138672
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,32,balanced,0.17297279834747314
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,32,balanced,0.29871039986610415
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,32,balanced,1.1310687899589538
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,16,balanced,0.1768864005804062
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,16,balanced,0.19942080080509186
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,16,balanced,1.118556809425354
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,16,balanced,1.0010016024112702
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,8,balanced,0.16790399849414825
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,8,balanced,0.17170239984989166
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,8,balanced,1.0777183771133423
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,8,balanced,0.9975264191627502
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,2,balanced,0.25367999970912936
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,2,balanced,0.734304004907608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,2,balanced,1.8813087940216064
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,32,balanced,0.1704255998134613
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,32,balanced,0.20945920050144196
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,32,balanced,1.1297247886657715
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,32,balanced,1.0099263906478881
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,4,balanced,0.24511360079050065
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,4,balanced,0.40750719904899596
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,4,balanced,1.6603455901145936
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,16,balanced,0.2549664005637169
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,16,balanced,0.7825663924217224
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,16,balanced,1.7508512020111084
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,8,balanced,0.24089280068874358
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,8,balanced,0.7233280003070831
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,8,balanced,1.6930751919746398
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,0.2524512052536011
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,0.7000415980815887
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,1.8797664046287537
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,2.8059423685073854
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,32,balanced,0.2484800025820732
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,32,balanced,0.8203776001930236
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,32,balanced,1.7828319907188415
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,4,balanced,0.2427423983812332
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,4,balanced,0.3691616028547287
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,4,balanced,1.6619680166244506
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,4,balanced,1.9191391944885254
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,16,balanced,0.2510207980871201
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,16,balanced,0.4270720034837723
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,16,balanced,1.7508640170097352
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,8,balanced,0.24341760128736495
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,8,balanced,0.4036671966314316
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,8,balanced,1.692192006111145
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,32,balanced,0.2438751995563507
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,32,balanced,0.45386239886283875
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,32,balanced,1.7807007908821106
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,2,balanced,0.4611968010663986
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,2,balanced,1.7980159878730775
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,2,balanced,3.825168013572693
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,16,balanced,0.2512256011366844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,16,balanced,0.29005120396614076
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,16,balanced,1.7511263966560364
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,16,balanced,1.5501312017440796
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,8,balanced,0.24187840074300765
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,8,balanced,0.24552319794893265
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,8,balanced,1.6931551933288573
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,8,balanced,1.5553279995918274
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,32,balanced,0.243955197930336
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,32,balanced,0.3085408002138138
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,32,balanced,1.7790655970573426
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,32,balanced,1.579967999458313
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,4,balanced,0.44439999759197235
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,4,balanced,1.3413600087165833
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,4,balanced,3.288700795173645
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,2,balanced,0.461132800579071
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,2,balanced,1.425276803970337
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,2,balanced,3.706339192390442
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,8,balanced,0.4359967976808548
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,8,balanced,1.3994783878326416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,8,balanced,3.3371872425079347
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,16,balanced,0.441772797703743
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,16,balanced,1.5182976007461548
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,16,balanced,3.450380825996399
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,32,balanced,0.4356416016817093
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,32,balanced,1.6095392227172851
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,32,balanced,3.531382441520691
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,4,balanced,0.44226239919662474
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,4,balanced,0.781007993221283
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,4,balanced,3.286073637008667
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,0.4621823996305466
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,1.3541696071624756
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,3.8385151863098144
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,5.551568031311035
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,8,balanced,0.44059199690818784
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,8,balanced,0.7751199901103973
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,8,balanced,3.3383296012878416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,16,balanced,0.44326399862766264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,16,balanced,0.815798407793045
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,16,balanced,3.4506976127624513
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,4,balanced,0.4451487988233566
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,4,balanced,0.7013855993747711
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,4,balanced,3.2846816062927244
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,4,balanced,3.8110336303710937
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,32,balanced,0.4384128004312515
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,32,balanced,0.8697248041629791
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,32,balanced,3.532096028327942
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,8,balanced,0.43981760144233706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,8,balanced,0.45815359652042387
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,8,balanced,3.338643217086792
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,8,balanced,3.037923216819763
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,16,balanced,0.44270400404930116
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,16,balanced,0.537961608171463
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,16,balanced,3.4504960060119627
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,16,balanced,3.0473439931869506
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,32,balanced,0.4385919988155365
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,32,balanced,0.5797088027000428
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,32,balanced,3.5318047761917115
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,32,balanced,3.108288025856018
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,48,balanced,0.07178240083158016
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,48,balanced,0.06345280036330223
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,48,balanced,0.07951039895415306
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,48,balanced,0.06943680047988891
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,48,balanced,0.057891200110316274
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,48,balanced,0.0760000005364418
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,48,balanced,0.07219840064644814
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,48,balanced,0.06071360036730766
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,48,balanced,0.07970560044050216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,48,balanced,0.08114560022950172
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,48,balanced,0.06617600023746491
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,48,balanced,0.05902079939842224
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,48,balanced,0.08154879957437515
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,48,balanced,0.06859200112521649
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,48,balanced,0.05682240016758442
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,48,balanced,0.076927999407053
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,48,balanced,0.06827199980616569
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,48,balanced,0.052319999784231186
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,48,balanced,0.07329599857330323
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,48,balanced,0.07921279966831207
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,48,balanced,0.06653439998626709
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,48,balanced,0.05086720027029514
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,48,balanced,0.07699839919805526
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,48,balanced,0.06943679936230182
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,48,balanced,0.052911999821662906
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,48,balanced,0.07668479904532433
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,48,balanced,0.06634559966623783
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,48,balanced,0.05190400034189224
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,48,balanced,0.07836799994111061
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,48,balanced,0.08059839978814125
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,48,balanced,0.0641407996416092
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,48,balanced,0.04779839999973774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,48,balanced,0.07561920061707497
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,48,balanced,0.06339200101792812
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,48,balanced,0.04946560002863407
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,48,balanced,0.07476800009608268
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,48,balanced,0.06427520029246807
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,48,balanced,0.051894400268793106
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,48,balanced,0.073785600066185
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,48,balanced,0.07701120004057885
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,48,balanced,0.06582400016486645
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,48,balanced,0.0576895996928215
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,48,balanced,0.07698559984564782
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,48,balanced,0.06183039993047714
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,48,balanced,0.05121919959783554
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,48,balanced,0.08012800067663192
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,48,balanced,0.06010559946298599
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,48,balanced,0.05161599963903427
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,48,balanced,0.08147839903831482
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,48,balanced,0.08087999969720841
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,48,balanced,0.06334399990737438
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,48,balanced,0.0555808000266552
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,48,balanced,0.08098559901118278
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,48,balanced,0.06130880080163479
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,48,balanced,0.05219199992716313
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,48,balanced,0.08267839848995209
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,48,balanced,0.06173120029270649
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,48,balanced,0.053545599803328514
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,48,balanced,0.08376960009336472
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,48,balanced,0.08501760140061379
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,48,balanced,0.06254080012440681
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,48,balanced,0.061375999823212624
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,48,balanced,0.080086400359869
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,48,balanced,0.05852159932255745
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,48,balanced,0.05072000026702881
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,48,balanced,0.08223359808325767
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,48,balanced,0.06097279973328114
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,48,balanced,0.05164800025522709
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,48,balanced,0.07860479950904846
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,48,balanced,0.07831680029630661
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,48,balanced,0.059702400118112564
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,48,balanced,0.058553599193692206
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,48,balanced,0.07765760049223899
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,48,balanced,0.06162559986114502
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,48,balanced,0.05206720046699047
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,48,balanced,0.07911040037870407
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,48,balanced,0.061644799262285235
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,48,balanced,0.051817599684000015
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,48,balanced,0.08211519941687584
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,48,balanced,0.0805792011320591
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,48,balanced,0.05895999930799008
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,48,balanced,0.055135999247431754
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,48,balanced,0.07795519977807999
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,48,balanced,0.06131840012967586
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,48,balanced,0.052243199571967124
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,48,balanced,0.07809600010514259
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,48,balanced,0.06094080023467541
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,48,balanced,0.052211199700832364
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,48,balanced,0.0832096017897129
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,48,balanced,0.08431039974093438
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,48,balanced,0.060860799998044966
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,48,balanced,0.06303360015153885
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,48,balanced,0.08071680068969726
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,48,balanced,0.06353279985487462
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,48,balanced,0.049878400191664694
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,48,balanced,0.08607999905943871
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,48,balanced,0.0605055995285511
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,48,balanced,0.051744000241160396
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,48,balanced,0.08390079960227012
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,48,balanced,0.08568640053272247
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,48,balanced,0.06779839992523193
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,48,balanced,0.06008000001311302
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,48,balanced,0.07816959917545319
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,48,balanced,0.060844799503684044
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,48,balanced,0.0536896001547575
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,48,balanced,0.08130239993333817
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,48,balanced,0.05855999998748303
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,48,balanced,0.05227839983999729
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,48,balanced,0.08180160000920296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,48,balanced,0.07852799966931343
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,48,balanced,0.060736000165343286
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,48,balanced,0.05860160067677498
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,48,balanced,0.08037440031766892
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,48,balanced,0.06054400019347668
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,48,balanced,0.05169280022382736
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,48,balanced,0.0775968000292778
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,48,balanced,0.058086400106549266
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,48,balanced,0.04814079999923706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,48,balanced,0.07948160022497178
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,48,balanced,0.0823295995593071
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,48,balanced,0.05760639980435371
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,48,balanced,0.06068800017237663
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,48,balanced,0.07953600063920022
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,48,balanced,0.06106239967048168
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,48,balanced,0.05286079905927181
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,48,balanced,0.09479680135846139
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,48,balanced,0.06338239945471287
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,48,balanced,0.05333760008215904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,48,balanced,0.07951679974794387
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,48,balanced,0.0778271995484829
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,48,balanced,0.060179199650883676
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,48,balanced,0.06876479908823967
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,48,balanced,0.08457919955253601
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,48,balanced,0.06315519995987415
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,48,balanced,0.052838399261236194
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,48,balanced,0.08128959983587265
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,48,balanced,0.0634591992944479
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,48,balanced,0.0807872001081705
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,48,balanced,0.08048000037670136
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,48,balanced,0.08194560036063195
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,48,balanced,0.060352000221610066
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,48,balanced,0.06747200004756451
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,48,balanced,0.08796480000019073
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,48,balanced,0.06921920031309128
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,48,balanced,0.0564863994717598
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,48,balanced,0.08530879989266396
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,48,balanced,0.06006720066070557
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,48,balanced,0.05099200047552586
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,48,balanced,0.08548799902200699
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,48,balanced,0.08592960014939308
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,48,balanced,0.06103999949991703
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,48,balanced,0.07310720086097718
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,48,balanced,0.0911360003054142
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,48,balanced,0.059631998836994174
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,48,balanced,0.05454719997942448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,48,balanced,0.09211520105600357
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,48,balanced,0.05960960052907467
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,48,balanced,0.057875200361013415
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,48,balanced,0.09276479929685592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,48,balanced,0.08592320010066032
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,48,balanced,0.06263039894402027
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,48,balanced,0.07913919985294342
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,48,balanced,0.10486719980835915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,48,balanced,0.06718720048666
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,48,balanced,0.06164799965918064
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,48,balanced,0.10441920012235642
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,48,balanced,0.0637695997953415
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,48,balanced,0.05654079988598824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,48,balanced,0.10555839985609054
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,48,balanced,0.09838079884648324
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,48,balanced,0.060892799869179726
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,48,balanced,0.09453119933605195
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,48,balanced,0.13415040001273154
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,48,balanced,0.07062719985842705
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,48,balanced,0.06962880082428455
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,48,balanced,0.13191999942064286
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,48,balanced,0.06357119977474213
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,48,balanced,0.06194240041077137
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,48,balanced,0.12978240102529526
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,48,balanced,0.11713280007243157
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,48,balanced,0.06444800049066543
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,48,balanced,0.11083199977874755
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,48,balanced,0.15821440368890763
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,48,balanced,0.06318720020353794
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,48,balanced,0.07814719974994659
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,48,balanced,0.15614400058984756
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,48,balanced,0.061273600161075595
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,48,balanced,0.07025600001215934
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,48,balanced,0.15461120009422302
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,48,balanced,0.14199040383100509
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,48,balanced,0.06829760037362576
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,48,balanced,0.1444928005337715
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,48,balanced,0.20905279964208603
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,48,balanced,0.063952000066638
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,48,balanced,0.09269439950585365
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,48,balanced,0.20413440018892287
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,48,balanced,0.06945599988102913
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,48,balanced,0.07781120017170906
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,48,balanced,0.20536959916353226
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,48,balanced,0.1844671994447708
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,48,balanced,0.06578240096569062
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,48,balanced,0.17771839797496797
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,48,balanced,0.2550527974963188
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,48,balanced,0.06771839894354344
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,48,balanced,0.11294720023870468
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,48,balanced,0.25669119954109193
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,48,balanced,0.06681279838085175
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,48,balanced,0.09000000059604644
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,48,balanced,0.2584960043430328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,48,balanced,0.22368000000715255
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,48,balanced,0.07273919954895973
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,48,balanced,0.24568640291690827
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,48,balanced,0.3532287985086441
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,48,balanced,0.07287999913096428
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,48,balanced,0.1458751991391182
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,48,balanced,0.35551680326461793
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,48,balanced,0.09766719937324524
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,48,balanced,0.11116159930825234
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,48,balanced,0.3513887971639633
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,48,balanced,0.3010560065507889
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,48,balanced,0.07760000079870225
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,48,balanced,0.3206400007009506
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,48,balanced,0.4558784008026123
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,48,balanced,0.09824960008263588
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,48,balanced,0.18153599947690963
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,48,balanced,0.46120000183582305
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,48,balanced,0.07906240001320838
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,48,balanced,0.1355103999376297
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,48,balanced,0.46039679944515227
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,48,balanced,0.3825632005929947
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,48,balanced,0.09174399971961975
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,48,balanced,0.46127359867095946
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,48,balanced,0.6552703857421875
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,48,balanced,0.09364479929208755
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,48,balanced,0.2550304040312767
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,48,balanced,0.6565920054912567
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,48,balanced,0.09136959910392761
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,48,balanced,0.1786848023533821
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,48,balanced,0.6595616042613983
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,48,balanced,0.5468191981315613
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,48,balanced,0.10253119990229606
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,48,balanced,0.6025120079517364
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,48,balanced,0.8608319938182831
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,48,balanced,0.10534399971365929
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,48,balanced,0.3253503978252411
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,48,balanced,0.868176007270813
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,48,balanced,0.10482879877090454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,48,balanced,0.2212799981236458
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,48,balanced,0.8618559956550598
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,48,balanced,0.7030655980110169
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,48,balanced,0.13304000049829484
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,48,balanced,0.8960224032402039
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,48,balanced,1.2727199912071228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,48,balanced,0.12999359965324403
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,48,balanced,0.4760447978973389
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,48,balanced,1.2627936005592346
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,48,balanced,0.1293791987001896
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,48,balanced,0.3104607999324799
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,48,balanced,1.268729603290558
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,48,balanced,1.0311039924621581
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,48,balanced,0.15096000134944915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,48,balanced,1.1905824065208435
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,48,balanced,1.666809606552124
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,48,balanced,0.15331519842147828
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,48,balanced,0.6255039930343628
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,48,balanced,1.681558406352997
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,48,balanced,0.16550400108098984
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,48,balanced,0.39732800126075746
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,48,balanced,1.6638975977897643
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,48,balanced,1.3460031867027282
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,48,balanced,0.17917119860649108
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,48,balanced,1.4867199897766112
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,48,balanced,2.090569591522217
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,48,balanced,0.18956480026245118
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,48,balanced,0.7833343982696533
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,48,balanced,2.0754143953323365
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,48,balanced,0.17866560071706772
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,48,balanced,0.4877247989177704
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,48,balanced,2.078441619873047
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,48,balanced,1.6765056133270264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,48,balanced,0.2563967972993851
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,48,balanced,2.3780192136764526
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,48,balanced,3.31703040599823
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,48,balanced,0.2535423994064331
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,48,balanced,1.2523391962051391
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,48,balanced,3.2925120115280153
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,48,balanced,0.2537663996219635
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,48,balanced,0.7615551948547363
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,48,balanced,3.3147648096084597
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,48,balanced,2.634995174407959
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,48,balanced,0.4564384013414383
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,48,balanced,4.8406623840332035
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,48,balanced,6.534041547775269
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,48,balanced,0.4615744024515152
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,48,balanced,2.519823980331421
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,48,balanced,6.528387212753296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,48,balanced,0.462281596660614
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,48,balanced,1.477232003211975
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,48,balanced,6.537779188156128
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,48,balanced,5.170265626907349
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,64,balanced,0.07527679949998856
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,64,balanced,0.05571519955992699
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,64,balanced,0.0826432004570961
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,64,balanced,0.0694047998636961
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,64,balanced,0.055119999870657924
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,64,balanced,0.07391999959945679
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,64,balanced,0.07138240076601506
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,64,balanced,0.05944960005581379
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,64,balanced,0.07499520033597946
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,64,balanced,0.07788480073213577
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,64,balanced,0.06659840010106563
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,64,balanced,0.05174719989299774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,64,balanced,0.07492799982428551
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,64,balanced,0.0661023996770382
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,64,balanced,0.05598400086164475
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,64,balanced,0.07742079943418503
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,64,balanced,0.06970559917390347
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,64,balanced,0.05389759950339794
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,64,balanced,0.07471680045127868
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,64,balanced,0.07603520005941392
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,64,balanced,0.06411520056426526
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,64,balanced,0.04962240047752857
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,64,balanced,0.07355839982628823
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,64,balanced,0.06671359986066819
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,64,balanced,0.04917760007083416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,64,balanced,0.07503359913825988
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,64,balanced,0.06441280022263526
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,64,balanced,0.07619839869439601
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,64,balanced,0.07688959911465645
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,64,balanced,0.0843552015721798
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,64,balanced,0.07395199984312058
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,64,balanced,0.0567391999065876
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,64,balanced,0.07679040059447288
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,64,balanced,0.07658239975571632
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,64,balanced,0.05835520066320896
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,64,balanced,0.0775456003844738
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,64,balanced,0.07470720075070858
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,64,balanced,0.057919999957084654
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,64,balanced,0.0748319998383522
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,64,balanced,0.08692800030112266
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,64,balanced,0.07470720037817954
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,64,balanced,0.06386879943311215
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,64,balanced,0.11028480231761932
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,64,balanced,0.07139520049095154
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,64,balanced,0.061820799857378005
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,64,balanced,0.1043967992067337
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,64,balanced,0.06782720051705837
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,64,balanced,0.06016639992594719
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,64,balanced,0.11044480055570602
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,64,balanced,0.10321600064635277
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,64,balanced,0.07284480035305023
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,64,balanced,0.06606080047786236
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,64,balanced,0.10399999991059303
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,64,balanced,0.07193280011415482
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,64,balanced,0.05679040029644966
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,64,balanced,0.10281920060515404
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,64,balanced,0.0735679991543293
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,64,balanced,0.06100480109453201
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,64,balanced,0.10760959908366204
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,64,balanced,0.1033503994345665
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,64,balanced,0.07028160095214844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,64,balanced,0.06845759861171245
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,64,balanced,0.1053983986377716
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,64,balanced,0.06960319876670837
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,64,balanced,0.05314239971339703
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,64,balanced,0.10094400122761726
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,64,balanced,0.07211200073361397
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,64,balanced,0.059600000455975535
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,64,balanced,0.09683840051293373
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,64,balanced,0.09894719943404198
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,64,balanced,0.0661631990224123
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,64,balanced,0.06434240005910397
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,64,balanced,0.09031680077314377
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,64,balanced,0.07157439962029458
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,64,balanced,0.05846719928085804
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,64,balanced,0.10679039955139161
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,64,balanced,0.07082879990339279
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,64,balanced,0.05911999978125095
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,64,balanced,0.10054399967193603
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,64,balanced,0.10131199806928634
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,64,balanced,0.06575999893248081
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,64,balanced,0.06169600039720535
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,64,balanced,0.09767360016703605
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,64,balanced,0.06472639963030816
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,64,balanced,0.05514879934489727
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,64,balanced,0.1032704010605812
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,64,balanced,0.06822720058262348
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,64,balanced,0.061580800265073773
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,64,balanced,0.10402880012989044
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,64,balanced,0.10623359978199005
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,64,balanced,0.06690240018069744
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,64,balanced,0.06383679993450642
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,64,balanced,0.10603839978575706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,64,balanced,0.06635520048439503
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,64,balanced,0.058368000015616414
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,64,balanced,0.10396159887313842
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,64,balanced,0.07569279968738556
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,64,balanced,0.056995199993252754
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,64,balanced,0.10459199920296669
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,64,balanced,0.10599040091037751
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,64,balanced,0.17963840551674365
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,64,balanced,0.07047680094838142
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,64,balanced,0.09468799903988838
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,64,balanced,0.06941119953989983
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,64,balanced,0.05777920000255108
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,64,balanced,0.10191680043935776
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,64,balanced,0.06754240095615387
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,64,balanced,0.056671999394893646
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,64,balanced,0.09782079979777336
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,64,balanced,0.09781120046973228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,64,balanced,0.06609920039772987
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,64,balanced,0.0722848005592823
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,64,balanced,0.10372160002589226
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,64,balanced,0.06768319979310036
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,64,balanced,0.05852480046451092
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,64,balanced,0.09653119966387749
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,64,balanced,0.06741440072655677
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,64,balanced,0.062822400406003
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,64,balanced,0.1011583998799324
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,64,balanced,0.09711039960384368
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,64,balanced,0.06515840031206607
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,64,balanced,0.07336000055074691
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,64,balanced,0.10435200035572052
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,64,balanced,0.0654848001897335
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,64,balanced,0.054784000292420386
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,64,balanced,0.10309119969606399
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,64,balanced,0.06943359971046448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,64,balanced,0.060953599959611894
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,64,balanced,0.09916800037026405
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,64,balanced,0.10121280029416084
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,64,balanced,0.06633280105888843
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,64,balanced,0.07091520093381405
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,64,balanced,0.10500160083174706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,64,balanced,0.06505599990487099
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,64,balanced,0.058342399448156355
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,64,balanced,0.09956160113215447
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,64,balanced,0.0661472000181675
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,64,balanced,0.05670720003545284
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,64,balanced,0.10004160031676293
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,64,balanced,0.10063679963350296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,64,balanced,0.06654719933867455
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,64,balanced,0.08345919921994209
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,64,balanced,0.10429119989275933
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,64,balanced,0.06649919934570789
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,64,balanced,0.06280000023543834
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,64,balanced,0.10476800054311752
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,64,balanced,0.06716479882597923
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,64,balanced,0.061484800279140474
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,64,balanced,0.10285120084881783
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,64,balanced,0.10311039984226227
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,64,balanced,0.06704960018396378
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,64,balanced,0.07938240095973015
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,64,balanced,0.09906559959053993
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,64,balanced,0.06682560183107852
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,64,balanced,0.06082880049943924
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,64,balanced,0.09876799881458283
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,64,balanced,0.07150719948112964
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,64,balanced,0.0624479990452528
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,64,balanced,0.09932479858398438
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,64,balanced,0.09099519997835159
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,64,balanced,0.06471999995410442
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,64,balanced,0.08099199905991554
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,64,balanced,0.10653440058231353
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,64,balanced,0.06466239988803864
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,64,balanced,0.06269439980387688
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,64,balanced,0.10356480032205581
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,64,balanced,0.06870399862527847
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,64,balanced,0.05998080000281334
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,64,balanced,0.10033280029892921
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,64,balanced,0.10382720082998276
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,64,balanced,0.06658239960670471
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,64,balanced,0.09695680066943169
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,64,balanced,0.1259328007698059
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,64,balanced,0.06531840078532695
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,64,balanced,0.07107200026512146
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,64,balanced,0.12179520055651664
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,64,balanced,0.07698880061507225
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,64,balanced,0.06698239967226982
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,64,balanced,0.12498240098357201
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,64,balanced,0.1156543992459774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,64,balanced,0.07516159936785698
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,64,balanced,0.10630720108747482
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,64,balanced,0.14728320240974427
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,64,balanced,0.07350080013275147
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,64,balanced,0.07643839940428734
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,64,balanced,0.14233599752187728
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,64,balanced,0.07131840139627457
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,64,balanced,0.06596160009503364
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,64,balanced,0.14285120218992234
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,64,balanced,0.13666239976882935
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,64,balanced,0.07597120180726051
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,64,balanced,0.13410559743642808
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,64,balanced,0.18749119937419892
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,64,balanced,0.07562239989638328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,64,balanced,0.08949119821190835
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,64,balanced,0.18343999832868577
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,64,balanced,0.07371839992702008
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,64,balanced,0.07561599984765052
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,64,balanced,0.18307520151138307
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,64,balanced,0.16916159987449647
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,64,balanced,0.0747423991560936
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,64,balanced,0.15740800201892852
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,64,balanced,0.2267199993133545
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,64,balanced,0.07459520064294338
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,64,balanced,0.1051936000585556
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,64,balanced,0.2229312002658844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,64,balanced,0.07599680051207543
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,64,balanced,0.08732799962162971
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,64,balanced,0.2264384001493454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,64,balanced,0.20477759689092637
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,64,balanced,0.07662720009684562
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,64,balanced,0.22539840191602706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,64,balanced,0.3072383999824524
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,64,balanced,0.08281920105218887
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,64,balanced,0.1313919998705387
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,64,balanced,0.3092672020196915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,64,balanced,0.07888640016317368
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,64,balanced,0.10360959991812706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,64,balanced,0.30643520355224607
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,64,balanced,0.27410240173339845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,64,balanced,0.08047360107302666
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,64,balanced,0.2588896006345749
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,64,balanced,0.38513919711112976
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,64,balanced,0.08348480015993118
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,64,balanced,0.15817920118570328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,64,balanced,0.3833855986595154
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,64,balanced,0.08241280093789101
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,64,balanced,0.11914240121841431
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,64,balanced,0.3840416014194489
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,64,balanced,0.3463039994239807
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,64,balanced,0.0963296003639698
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,64,balanced,0.37189760208129885
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,64,balanced,0.5576704025268555
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,64,balanced,0.09415040016174317
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,64,balanced,0.2131936013698578
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,64,balanced,0.5537952005863189
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,64,balanced,0.09284799993038177
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,64,balanced,0.15415679961442946
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,64,balanced,0.5537440001964569
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,64,balanced,0.48008640110492706
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,64,balanced,0.10466560125350952
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,64,balanced,0.48770560026168824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,64,balanced,0.7335871994495392
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,64,balanced,0.1064191997051239
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,64,balanced,0.2671071976423264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,64,balanced,0.7323871970176696
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,64,balanced,0.108143999427557
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,64,balanced,0.188646399974823
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,64,balanced,0.7298528015613556
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,64,balanced,0.6330591917037964
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,64,balanced,0.1334079995751381
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,64,balanced,0.7124127924442292
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,64,balanced,1.0706911921501159
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,64,balanced,0.13210239857435227
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,64,balanced,0.37911359667778016
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,64,balanced,1.07298241853714
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,64,balanced,0.13045439943671228
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,64,balanced,0.2590847983956337
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,64,balanced,1.0724639892578125
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,64,balanced,0.9016831994056702
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,64,balanced,0.15570240020751952
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,64,balanced,0.9407296001911163
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,64,balanced,1.4208096027374268
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,64,balanced,0.15858560055494308
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,64,balanced,0.4923680007457733
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,64,balanced,1.4165696144104003
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,64,balanced,0.1584223985671997
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,64,balanced,0.3256160020828247
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,64,balanced,1.4178591966629028
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,64,balanced,1.178163194656372
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,64,balanced,0.1857023999094963
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,64,balanced,1.172432005405426
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,64,balanced,1.7712448000907899
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,64,balanced,0.18370880037546158
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,64,balanced,0.6180607974529266
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,64,balanced,1.7705055952072144
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,64,balanced,0.1864064007997513
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,64,balanced,0.39648639857769014
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,64,balanced,1.764784002304077
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,64,balanced,1.4694944262504577
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,64,balanced,0.2569856017827988
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,64,balanced,1.8736543893814086
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,64,balanced,2.8128480195999144
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,64,balanced,0.25380159467458724
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,64,balanced,0.9715743958950043
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,64,balanced,2.8169311761856077
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,64,balanced,0.2503551974892616
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,64,balanced,0.5994335949420929
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,64,balanced,2.817574381828308
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,64,balanced,2.3170048236846923
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,64,balanced,0.4554816037416458
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,64,balanced,3.744444799423218
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,64,balanced,5.645920038223267
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,64,balanced,0.4612256020307541
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,64,balanced,1.9498560070991515
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,64,balanced,5.640854406356811
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,64,balanced,0.4635712027549744
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,64,balanced,1.1379744052886962
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,64,balanced,5.644374370574951
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,64,balanced,4.569228744506836
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,72,balanced,0.08456639871001244
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,72,balanced,0.06690559983253479
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,72,balanced,0.08445120081305504
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,72,balanced,0.08342719972133636
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,72,balanced,0.06369599997997284
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,72,balanced,0.08287360072135926
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,72,balanced,0.08010239899158478
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,72,balanced,0.062412799522280696
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,72,balanced,0.08464639931917191
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,72,balanced,0.08592639937996864
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,72,balanced,0.07695679962635041
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,72,balanced,0.05760639980435371
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,72,balanced,0.0820095993578434
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,72,balanced,0.07682560011744499
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,72,balanced,0.06302400082349777
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,72,balanced,0.08376320004463196
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,72,balanced,0.079721599817276
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,72,balanced,0.06271039992570877
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,72,balanced,0.08102079927921295
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,72,balanced,0.08219520002603531
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,72,balanced,0.0780000001192093
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,72,balanced,0.060630399361252786
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,72,balanced,0.08284160047769547
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,72,balanced,0.0818496011197567
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,72,balanced,0.0620448000729084
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,72,balanced,0.08164159953594208
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,72,balanced,0.07544639967381954
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,72,balanced,0.06234560012817383
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,72,balanced,0.08066879883408547
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,72,balanced,0.07841279953718186
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,72,balanced,0.07582719996571541
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,72,balanced,0.06097599901258945
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,72,balanced,0.08584320023655892
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,72,balanced,0.0757311999797821
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,72,balanced,0.06194559969007969
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,72,balanced,0.08299200087785721
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,72,balanced,0.07731840088963508
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,72,balanced,0.061062400415539744
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,72,balanced,0.08080320134758949
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,72,balanced,0.09059199988842011
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,72,balanced,0.07110080048441887
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,72,balanced,0.09793920069932938
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,72,balanced,0.12393280118703842
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,72,balanced,0.0824960008263588
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,72,balanced,0.06798399984836578
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,72,balanced,0.11521920040249825
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,72,balanced,0.08530559986829758
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,72,balanced,0.06865920089185237
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,72,balanced,0.11415039896965026
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,72,balanced,0.10977279916405677
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,72,balanced,0.0790560007095337
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,72,balanced,0.07919680029153824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,72,balanced,0.11161280050873756
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,72,balanced,0.08232639878988265
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,72,balanced,0.07182080037891865
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,72,balanced,0.11389440074563026
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,72,balanced,0.08090879879891873
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,72,balanced,0.06863359995186329
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,72,balanced,0.11498880088329315
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,72,balanced,0.11154880002140999
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,72,balanced,0.0824512004852295
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,72,balanced,0.07557120025157929
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,72,balanced,0.11495999991893768
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,72,balanced,0.07825279980897903
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,72,balanced,0.07347840033471584
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,72,balanced,0.1159328006207943
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,72,balanced,0.07992319986224175
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,72,balanced,0.1019968006759882
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,72,balanced,0.11683840081095695
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,72,balanced,0.10784000009298325
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,72,balanced,0.07839359976351261
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,72,balanced,0.07744319923222065
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,72,balanced,0.11696639955043793
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,72,balanced,0.0773376002907753
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,72,balanced,0.07029120028018951
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,72,balanced,0.11350080147385597
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,72,balanced,0.07758720070123673
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,72,balanced,0.07172480002045631
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,72,balanced,0.114444799721241
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,72,balanced,0.10782719850540161
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,72,balanced,0.08035840094089508
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,72,balanced,0.08134080097079277
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,72,balanced,0.12176320105791091
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,72,balanced,0.07757119983434677
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,72,balanced,0.06841920055449009
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,72,balanced,0.11695359945297241
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,72,balanced,0.08167039826512337
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,72,balanced,0.06808320060372353
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,72,balanced,0.11234560012817382
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,72,balanced,0.11226879879832267
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,72,balanced,0.08033600077033043
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,72,balanced,0.08262080028653145
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,72,balanced,0.11897600144147873
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,72,balanced,0.08129919916391373
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,72,balanced,0.0716767992824316
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,72,balanced,0.11749760136008262
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,72,balanced,0.08445120006799697
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,72,balanced,0.07357760034501552
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,72,balanced,0.11226880177855492
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,72,balanced,0.11739519983530045
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,72,balanced,0.0779231995344162
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,72,balanced,0.08125439882278443
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,72,balanced,0.11326080113649369
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,72,balanced,0.07970559969544411
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,72,balanced,0.07129279933869839
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,72,balanced,0.11185600087046624
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,72,balanced,0.081020800024271
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,72,balanced,0.06902080029249191
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,72,balanced,0.1176352009177208
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,72,balanced,0.11694400086998939
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,72,balanced,0.08504639938473701
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,72,balanced,0.08345920071005822
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,72,balanced,0.11625280007719993
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,72,balanced,0.08461119905114174
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,72,balanced,0.07174720056355
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,72,balanced,0.11426240056753159
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,72,balanced,0.0754720002412796
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,72,balanced,0.07054079994559288
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,72,balanced,0.11672320067882538
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,72,balanced,0.11157759949564934
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,72,balanced,0.08112000003457069
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,72,balanced,0.08583999797701836
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,72,balanced,0.1819264017045498
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,72,balanced,0.0795840010046959
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,72,balanced,0.06928640007972717
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,72,balanced,0.11870080009102821
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,72,balanced,0.08222079873085023
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,72,balanced,0.07310400009155274
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,72,balanced,0.11214080005884171
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,72,balanced,0.1161920003592968
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,72,balanced,0.07972479984164238
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,72,balanced,0.08605439960956573
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,72,balanced,0.12187520116567611
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,72,balanced,0.07816959992051124
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,72,balanced,0.07280640080571174
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,72,balanced,0.11591679975390434
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,72,balanced,0.0885567992925644
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,72,balanced,0.0753024011850357
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,72,balanced,0.1156543992459774
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,72,balanced,0.11274240016937256
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,72,balanced,0.08496320024132728
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,72,balanced,0.09200320094823837
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,72,balanced,0.12141759991645813
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,72,balanced,0.08188480138778687
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,72,balanced,0.07521920055150985
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,72,balanced,0.11919359937310219
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,72,balanced,0.08289600014686585
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,72,balanced,0.08318079933524132
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,72,balanced,0.11791040077805519
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,72,balanced,0.11786880046129226
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,72,balanced,0.08168639913201332
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,72,balanced,0.09271999970078468
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,72,balanced,0.11974719911813736
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,72,balanced,0.08310719877481461
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,72,balanced,0.07474559992551803
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,72,balanced,0.12229439988732338
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,72,balanced,0.08109759986400604
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,72,balanced,0.0788223996758461
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,72,balanced,0.12035840153694152
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,72,balanced,0.11388159841299057
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,72,balanced,0.08246400058269501
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,72,balanced,0.10179520174860954
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,72,balanced,0.13892479985952377
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,72,balanced,0.08159999921917915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,72,balanced,0.08027519881725312
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,72,balanced,0.13837119936943054
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,72,balanced,0.08444159999489784
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,72,balanced,0.0793279990553856
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,72,balanced,0.13978880047798156
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,72,balanced,0.1229984000325203
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,72,balanced,0.08598719984292984
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,72,balanced,0.12320640012621879
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,72,balanced,0.16798079907894134
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,72,balanced,0.08248639926314354
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,72,balanced,0.08925439938902854
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,72,balanced,0.16972479969263077
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,72,balanced,0.08012159988284111
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,72,balanced,0.08066239953041077
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,72,balanced,0.16650240123271942
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,72,balanced,0.14846079796552658
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,72,balanced,0.08400959894061089
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,72,balanced,0.14989120215177537
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,72,balanced,0.20346559882164
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,72,balanced,0.08369600027799606
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,72,balanced,0.10029760003089905
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,72,balanced,0.2044863998889923
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,72,balanced,0.08369279950857163
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,72,balanced,0.08775360062718392
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,72,balanced,0.20577919781208037
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,72,balanced,0.17847039997577668
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,72,balanced,0.08239359855651855
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,72,balanced,0.201459202170372
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,72,balanced,0.2743200033903122
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,72,balanced,0.09032640084624291
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,72,balanced,0.12146240100264549
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,72,balanced,0.2730783998966217
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,72,balanced,0.08995840176939965
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,72,balanced,0.10399999916553497
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,72,balanced,0.2699072062969208
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,72,balanced,0.22873920053243638
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,72,balanced,0.08475200012326241
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,72,balanced,0.26025919914245604
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,72,balanced,0.3409855991601944
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,72,balanced,0.08704639971256256
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,72,balanced,0.14999999850988388
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,72,balanced,0.3426784038543701
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,72,balanced,0.09283840060234069
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,72,balanced,0.12520639970898628
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,72,balanced,0.34597760140895845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,72,balanced,0.28721280097961427
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,72,balanced,0.0903328001499176
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,72,balanced,0.37423679828643797
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,72,balanced,0.4820063978433609
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,72,balanced,0.09431679993867874
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,72,balanced,0.20309759825468063
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,72,balanced,0.48464320302009584
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,72,balanced,0.09809599965810775
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,72,balanced,0.15476159900426864
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,72,balanced,0.4847648024559021
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,72,balanced,0.3947839975357056
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,72,balanced,0.09527360051870346
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,72,balanced,0.49390400350093844
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,72,balanced,0.6295999944210052
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,72,balanced,0.09360000044107437
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,72,balanced,0.2609888017177582
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,72,balanced,0.6302528083324432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,72,balanced,0.09394880086183548
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,72,balanced,0.1870592013001442
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,72,balanced,0.62778240442276
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,72,balanced,0.5006336033344269
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,72,balanced,0.10700160041451454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,72,balanced,0.7393920004367829
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,72,balanced,0.9295871973037719
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,72,balanced,0.1067552000284195
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,72,balanced,0.37768320441246034
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,72,balanced,0.931193608045578
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,72,balanced,0.10833920016884804
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,72,balanced,0.2540480047464371
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,72,balanced,0.9284255981445313
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,72,balanced,0.721587210893631
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,72,balanced,0.11823359951376915
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,72,balanced,0.9904287993907929
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,72,balanced,1.2296447992324828
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,72,balanced,0.1215519979596138
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,72,balanced,0.4924032062292099
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,72,balanced,1.2279583811759949
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,72,balanced,0.11970560029149055
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,72,balanced,0.3264575958251953
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,72,balanced,1.2325984001159669
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,72,balanced,0.9369279980659485
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,72,balanced,0.14037439972162247
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,72,balanced,1.4940096020698548
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,72,balanced,1.8295040130615234
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,72,balanced,0.14344960153102876
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,72,balanced,0.7381248056888581
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,72,balanced,1.8309183955192565
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,72,balanced,0.14396799951791764
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,72,balanced,0.4688960015773773
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,72,balanced,1.8277887940406798
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,72,balanced,1.388047993183136
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,72,balanced,0.16779199838638306
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,72,balanced,2.0269440412521362
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,72,balanced,2.433772826194763
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,72,balanced,0.16629120111465454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,72,balanced,0.9922880113124848
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,72,balanced,2.4281951904296877
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,72,balanced,0.16393279880285264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,72,balanced,0.614351999759674
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,72,balanced,2.4342528104782106
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,72,balanced,1.8209631800651551
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,72,balanced,0.1924032002687454
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,72,balanced,2.5559871912002565
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,72,balanced,3.0416255950927735
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,72,balanced,0.19380159974098204
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,72,balanced,1.2694880008697509
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,72,balanced,3.055468797683716
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,72,balanced,0.1943135991692543
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,72,balanced,0.7502080023288726
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,72,balanced,3.044003200531006
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,72,balanced,2.262227249145508
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,72,balanced,0.2704768031835556
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,72,balanced,4.117555236816406
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,72,balanced,4.861487960815429
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,72,balanced,0.2658367961645126
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,72,balanced,2.0986047983169556
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,72,balanced,4.855260753631592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,72,balanced,0.26983679831027985
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,72,balanced,1.1749759912490845
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,72,balanced,4.861916732788086
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,72,balanced,3.5754080533981325
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,72,balanced,0.46169919669628146
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,72,balanced,8.375702476501464
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,72,balanced,9.69835205078125
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,72,balanced,0.46228799819946287
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,72,balanced,4.29211196899414
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,72,balanced,9.696246337890624
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,72,balanced,0.46026560068130495
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,72,balanced,2.360585618019104
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,72,balanced,9.707542419433594
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,72,balanced,7.02664966583252
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1,7168,8,256,2,balanced,0.0305104
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1,7168,8,256,2,balanced,0.025272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1,7168,8,256,2,balanced,0.037488
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1,7168,8,256,2,balanced,0.0292192
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1,7168,8,256,2,balanced,0.0236336
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1,7168,8,256,2,balanced,0.0393712
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.0308096
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.0256272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.0384576
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1,7168,8,256,2,balanced,0.038352
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2,7168,8,256,2,balanced,0.0291824
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2,7168,8,256,2,balanced,0.0238144
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2,7168,8,256,2,balanced,0.0377856
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2,7168,8,256,2,balanced,0.0311424
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2,7168,8,256,2,balanced,0.0246032
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2,7168,8,256,2,balanced,0.0376784
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.0287328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.0231792
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.0365216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2,7168,8,256,2,balanced,0.0369856
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4,7168,8,256,2,balanced,0.0280432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4,7168,8,256,2,balanced,0.0225536
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4,7168,8,256,2,balanced,0.037776
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4,7168,8,256,2,balanced,0.0283648
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4,7168,8,256,2,balanced,0.0215632
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4,7168,8,256,2,balanced,0.0380288
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.0287888
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.0239088
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.0372912
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4,7168,8,256,2,balanced,0.0374848
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8,7168,8,256,2,balanced,0.0291264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8,7168,8,256,2,balanced,0.0235616
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8,7168,8,256,2,balanced,0.0376832
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8,7168,8,256,2,balanced,0.0274112
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8,7168,8,256,2,balanced,0.0224688
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8,7168,8,256,2,balanced,0.0362768
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.0281856
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.0229184
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.0353552
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8,7168,8,256,2,balanced,0.0362016
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16,7168,8,256,2,balanced,0.0252
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16,7168,8,256,2,balanced,0.0231408
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16,7168,8,256,2,balanced,0.03732
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16,7168,8,256,2,balanced,0.0278032
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16,7168,8,256,2,balanced,0.0218112
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16,7168,8,256,2,balanced,0.0360944
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.0256304
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.0228512
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.0358416
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16,7168,8,256,2,balanced,0.03692
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32,7168,8,256,2,balanced,0.0275856
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32,7168,8,256,2,balanced,0.0219296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32,7168,8,256,2,balanced,0.0358288
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32,7168,8,256,2,balanced,0.0265952
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32,7168,8,256,2,balanced,0.021256
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32,7168,8,256,2,balanced,0.0347456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.027272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.0232512
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.036296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32,7168,8,256,2,balanced,0.0356752
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,48,7168,8,256,2,balanced,0.0272816
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,48,7168,8,256,2,balanced,0.0212944
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,48,7168,8,256,2,balanced,0.034848
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,48,7168,8,256,2,balanced,0.0270032
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,48,7168,8,256,2,balanced,0.0222896
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,48,7168,8,256,2,balanced,0.0354976
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.0254448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.0233184
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.0344608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,48,7168,8,256,2,balanced,0.0353568
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,64,7168,8,256,2,balanced,0.0270816
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,64,7168,8,256,2,balanced,0.0220496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,64,7168,8,256,2,balanced,0.035512
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,64,7168,8,256,2,balanced,0.0260672
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,64,7168,8,256,2,balanced,0.0223632
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,64,7168,8,256,2,balanced,0.0339184
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.026832
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.0228464
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.0343712
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,64,7168,8,256,2,balanced,0.0349456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,80,7168,8,256,2,balanced,0.026464
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,80,7168,8,256,2,balanced,0.0222352
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,80,7168,8,256,2,balanced,0.0342176
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,80,7168,8,256,2,balanced,0.027112
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,80,7168,8,256,2,balanced,0.0207152
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,80,7168,8,256,2,balanced,0.0351136
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.027544
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.0224592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.0371984
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,80,7168,8,256,2,balanced,0.0359328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,96,7168,8,256,2,balanced,0.0267376
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,96,7168,8,256,2,balanced,0.0214368
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,96,7168,8,256,2,balanced,0.033352
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,96,7168,8,256,2,balanced,0.0254448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,96,7168,8,256,2,balanced,0.0222704
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,96,7168,8,256,2,balanced,0.0343744
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.0255904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.0220752
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.03248
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,96,7168,8,256,2,balanced,0.036592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,128,7168,8,256,2,balanced,0.0262224
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,128,7168,8,256,2,balanced,0.021176
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,128,7168,8,256,2,balanced,0.0360048
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,128,7168,8,256,2,balanced,0.0254
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,128,7168,8,256,2,balanced,0.0195296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,128,7168,8,256,2,balanced,0.034144
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.02692
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.0224192
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.0342368
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,128,7168,8,256,2,balanced,0.03536
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,160,7168,8,256,2,balanced,0.0267808
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,160,7168,8,256,2,balanced,0.0216272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,160,7168,8,256,2,balanced,0.0350496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,160,7168,8,256,2,balanced,0.0248944
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,160,7168,8,256,2,balanced,0.02084
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,160,7168,8,256,2,balanced,0.034784
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.0266704
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.021872
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.0341344
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,160,7168,8,256,2,balanced,0.035832
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,192,7168,8,256,2,balanced,0.0259904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,192,7168,8,256,2,balanced,0.0213248
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,192,7168,8,256,2,balanced,0.0351616
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,192,7168,8,256,2,balanced,0.0270288
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,192,7168,8,256,2,balanced,0.020792
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,192,7168,8,256,2,balanced,0.034992
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.0265056
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.0207904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.0337024
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,192,7168,8,256,2,balanced,0.0348448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,256,7168,8,256,2,balanced,0.025376
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,256,7168,8,256,2,balanced,0.0214432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,256,7168,8,256,2,balanced,0.050448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,256,7168,8,256,2,balanced,0.0270096
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,256,7168,8,256,2,balanced,0.0202592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,256,7168,8,256,2,balanced,0.033912
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.02624
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.0228656
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.0329728
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,256,7168,8,256,2,balanced,0.0348096
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,320,7168,8,256,2,balanced,0.0267168
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,320,7168,8,256,2,balanced,0.0208784
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,320,7168,8,256,2,balanced,0.0338608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,320,7168,8,256,2,balanced,0.02576
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,320,7168,8,256,2,balanced,0.020464
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,320,7168,8,256,2,balanced,0.0344608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.026224
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.0219152
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.0339328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,320,7168,8,256,2,balanced,0.036432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,384,7168,8,256,2,balanced,0.02672
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,384,7168,8,256,2,balanced,0.0219648
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,384,7168,8,256,2,balanced,0.0350096
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,384,7168,8,256,2,balanced,0.0257264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,384,7168,8,256,2,balanced,0.020704
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,384,7168,8,256,2,balanced,0.0339744
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.0259008
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.022152
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.0342272
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,384,7168,8,256,2,balanced,0.0358448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,512,7168,8,256,2,balanced,0.0250976
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,512,7168,8,256,2,balanced,0.0244192
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,512,7168,8,256,2,balanced,0.0365488
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,512,7168,8,256,2,balanced,0.0271888
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,512,7168,8,256,2,balanced,0.0219616
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,512,7168,8,256,2,balanced,0.035304
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.0279424
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.0214496
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.0348432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,512,7168,8,256,2,balanced,0.0387584
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,768,7168,8,256,2,balanced,0.0253552
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,768,7168,8,256,2,balanced,0.025096
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,768,7168,8,256,2,balanced,0.0410368
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,768,7168,8,256,2,balanced,0.0284064
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,768,7168,8,256,2,balanced,0.0213072
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,768,7168,8,256,2,balanced,0.0401616
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.0263072
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.0233216
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.0399072
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,768,7168,8,256,2,balanced,0.046872
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1024,7168,8,256,2,balanced,0.0264128
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1024,7168,8,256,2,balanced,0.0276592
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1024,7168,8,256,2,balanced,0.0467408
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1024,7168,8,256,2,balanced,0.0260208
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1024,7168,8,256,2,balanced,0.0226752
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1024,7168,8,256,2,balanced,0.0467296
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.024576
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.023032
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.0464608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1024,7168,8,256,2,balanced,0.0543936
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,1536,7168,8,256,2,balanced,0.0270528
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,1536,7168,8,256,2,balanced,0.0333392
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,1536,7168,8,256,2,balanced,0.0592816
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,1536,7168,8,256,2,balanced,0.025712
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,1536,7168,8,256,2,balanced,0.0247264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,1536,7168,8,256,2,balanced,0.0595872
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.0271104
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.0260528
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.0591536
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,1536,7168,8,256,2,balanced,0.068896
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,2048,7168,8,256,2,balanced,0.0269632
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,2048,7168,8,256,2,balanced,0.0378064
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,2048,7168,8,256,2,balanced,0.0723776
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,2048,7168,8,256,2,balanced,0.0269328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,2048,7168,8,256,2,balanced,0.027448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,2048,7168,8,256,2,balanced,0.0717248
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.0273104
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.0287248
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.0714
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,2048,7168,8,256,2,balanced,0.0825904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,3072,7168,8,256,2,balanced,0.030448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,3072,7168,8,256,2,balanced,0.0476928
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,3072,7168,8,256,2,balanced,0.0981328
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,3072,7168,8,256,2,balanced,0.0295568
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,3072,7168,8,256,2,balanced,0.0327792
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,3072,7168,8,256,2,balanced,0.0983904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.0294384
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.0339568
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.0972368
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,3072,7168,8,256,2,balanced,0.112341
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,4096,7168,8,256,2,balanced,0.0335904
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,4096,7168,8,256,2,balanced,0.0573376
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,4096,7168,8,256,2,balanced,0.122691
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,4096,7168,8,256,2,balanced,0.0335104
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,4096,7168,8,256,2,balanced,0.0400352
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,4096,7168,8,256,2,balanced,0.122266
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.0330688
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.0380048
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.122451
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,4096,7168,8,256,2,balanced,0.140374
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,6144,7168,8,256,2,balanced,0.0397984
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,6144,7168,8,256,2,balanced,0.079192
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,6144,7168,8,256,2,balanced,0.172168
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,6144,7168,8,256,2,balanced,0.0401872
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,6144,7168,8,256,2,balanced,0.0545408
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,6144,7168,8,256,2,balanced,0.172542
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.0397504
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.0484656
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.171501
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,6144,7168,8,256,2,balanced,0.200701
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,8192,7168,8,256,2,balanced,0.0459888
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,8192,7168,8,256,2,balanced,0.099448
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,8192,7168,8,256,2,balanced,0.223941
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,8192,7168,8,256,2,balanced,0.0465264
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,8192,7168,8,256,2,balanced,0.0639408
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,8192,7168,8,256,2,balanced,0.221941
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.045456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.0594336
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.223154
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,8192,7168,8,256,2,balanced,0.257237
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,12288,7168,8,256,2,balanced,0.0587232
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,12288,7168,8,256,2,balanced,0.137552
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,12288,7168,8,256,2,balanced,0.32365
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,12288,7168,8,256,2,balanced,0.057968
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,12288,7168,8,256,2,balanced,0.0867376
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,12288,7168,8,256,2,balanced,0.322928
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,0.0590816
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,0.0808608
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,0.323226
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,12288,7168,8,256,2,balanced,0.37471
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,16384,7168,8,256,2,balanced,0.0715088
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,16384,7168,8,256,2,balanced,0.181794
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,16384,7168,8,256,2,balanced,0.424976
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,16384,7168,8,256,2,balanced,0.0702656
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,16384,7168,8,256,2,balanced,0.110432
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,16384,7168,8,256,2,balanced,0.42269
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,0.0698928
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,0.101408
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,0.424413
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,16384,7168,8,256,2,balanced,0.489944
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,20480,7168,8,256,2,balanced,0.0829728
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,20480,7168,8,256,2,balanced,0.219797
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,20480,7168,8,256,2,balanced,0.52816
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,20480,7168,8,256,2,balanced,0.0826528
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,20480,7168,8,256,2,balanced,0.13332
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,20480,7168,8,256,2,balanced,0.526926
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,0.082456
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,0.12312
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,0.528197
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,20480,7168,8,256,2,balanced,0.611245
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,32768,7168,8,256,2,balanced,0.121102
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,32768,7168,8,256,2,balanced,0.343546
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,32768,7168,8,256,2,balanced,0.83048
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,32768,7168,8,256,2,balanced,0.122557
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,32768,7168,8,256,2,balanced,0.203754
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,32768,7168,8,256,2,balanced,0.830173
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,0.121371
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,0.184581
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,0.830984
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,32768,7168,8,256,2,balanced,0.95957
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,float16,65536,7168,8,256,2,balanced,0.2222
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,float16,65536,7168,8,256,2,balanced,0.67068
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,float16,65536,7168,8,256,2,balanced,1.64435
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,fp8,65536,7168,8,256,2,balanced,0.221131
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,fp8,65536,7168,8,256,2,balanced,0.390504
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,fp8,65536,7168,8,256,2,balanced,1.64304
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_prepare,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,0.222574
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_dispatch,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,0.350693
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,1.64234
TRTLLM,1.2.0rc6,NVIDIA GB200,alltoall_combine_low_precision,MnnvlMoe,nvfp4,65536,7168,8,256,2,balanced,1.90552
