framework,version,device,op_name,kernel_source,batch_size,isl,num_heads,num_key_value_heads,head_dim,beam_width,attn_dtype,kv_cache_dtype,step,latency
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,0,0.02164799968401591
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,0,0.01952533299724261
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,0,0.025573333104451496
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,0,0.02784000088771184
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,64,64,128,1,float16,float16,0,0.10011733571688335
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,64,64,128,1,float16,fp8,0,0.07614399989446004
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,0,0.015146666516860327
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,0,0.018981333822011948
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,0,0.014912000546852747
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,0,0.01718933383623759
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,64,64,128,1,float16,float16,0,0.058101331194241844
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,64,64,128,1,float16,fp8,0,0.04204266766707102
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,0,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,0,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,0,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,64,64,128,1,float16,float16,0,0.029466666281223297
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,64,64,128,1,float16,fp8,0,0.027727998793125153
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,0,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,0,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,0,0.011312000453472137
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,0,0.011328000575304031
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,0,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,64,64,128,1,float16,float16,0,0.01939733326435089
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,64,64,128,1,float16,fp8,0,0.021322667598724365
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,0,0.009488000224033991
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,0,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,0,0.010474666953086853
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,0,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,64,64,128,1,float16,fp8,0,0.01714133347074191
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,64,64,128,1,float16,float16,0,0.015461333096027374
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,0,0.011146667102972666
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,0,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,0,0.010522666076819101
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,64,64,128,1,float16,fp8,0,0.015237333873907724
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,64,64,128,1,float16,float16,0,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,0,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,0,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,0,0.010842667271693548
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,0,0.00927466650803884
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,64,64,128,1,float16,fp8,0,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,64,64,128,1,float16,float16,0,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,0,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,0,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,0,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,0,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,0,0.01002133327225844
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,0,0.00960533320903778
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,0,0.009445333232482275
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,64,64,128,1,float16,float16,0,0.011136000355084738
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,64,64,128,1,float16,fp8,0,0.013264000415802002
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,0,0.009328000247478485
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,0,0.010650667051474253
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,0,0.021327999730904896
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,0,0.019088000059127808
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,0,0.025466665625572205
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,0,0.027482666075229645
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,48,48,128,1,float16,float16,0,0.0784800002972285
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,48,48,128,1,float16,fp8,0,0.05902933577696482
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,0,0.015263999501864115
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,0,0.014858666807413101
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,0,0.018415999909241993
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,0,0.017360000560681026
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,48,48,128,1,float16,float16,0,0.04387733340263367
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,48,48,128,1,float16,fp8,0,0.03341866781314214
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,0,0.012821332861979803
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,0,0.01314666618903478
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,0,0.014538666854302088
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,0,0.013189333180586496
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,48,48,128,1,float16,float16,0,0.023205332458019257
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,48,48,128,1,float16,fp8,0,0.023520000278949738
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,0,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,0,0.011472000430027643
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,0,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,48,48,128,1,float16,float16,0,0.01739199956258138
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,48,48,128,1,float16,fp8,0,0.01743999992807706
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,0,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,0,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,0,0.009797333429257074
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,0,0.011109333485364914
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,48,48,128,1,float16,fp8,0,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,48,48,128,1,float16,float16,0,0.013274667163689932
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,0,0.009674666449427605
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,0,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,0,0.010703999549150467
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,48,48,128,1,float16,fp8,0,0.01332266628742218
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,48,48,128,1,float16,float16,0,0.012826666235923767
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,0,0.009743999689817429
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,0,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,0,0.009519999846816063
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,0,0.010133333504199982
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,48,48,128,1,float16,float16,0,0.012597333639860153
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,48,48,128,1,float16,fp8,0,0.012970666090647379
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,0,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,0,0.009648000200589498
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,0,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,0,0.010442666709423065
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,0,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,48,48,128,1,float16,fp8,0,0.012906666845083237
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,48,48,128,1,float16,float16,0,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,0,0.009717333440979322
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,0,0.009893333539366722
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,0,0.009461333354314169
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,0,0.009648000200589498
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,0,0.021290667355060577
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,0,0.019493332753578823
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,0,0.025061334172884624
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,0,0.027717334528764088
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,40,40,128,1,float16,float16,0,0.06910933554172516
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,40,40,128,1,float16,fp8,0,0.049653331438700356
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,0,0.01586666703224182
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,0,0.015002666662136713
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,0,0.019082666685183842
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,0,0.016895999511082966
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,40,40,128,1,float16,float16,0,0.03482666611671448
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,40,40,128,1,float16,fp8,0,0.029290666182835896
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,0,0.012928000340859095
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,0,0.013066666821638743
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,0,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,0,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,40,40,128,1,float16,float16,0,0.02139200021823247
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,40,40,128,1,float16,fp8,0,0.02123733361562093
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,0,0.01119999960064888
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,0,0.011338666081428528
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,0,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,0,0.011157333850860596
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,40,40,128,1,float16,float16,0,0.016917333006858826
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,40,40,128,1,float16,fp8,0,0.0170666662355264
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,0,0.009626666704813639
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,0,0.010629333555698395
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,0,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,40,40,128,1,float16,float16,0,0.013370666652917862
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,40,40,128,1,float16,fp8,0,0.015002666662136713
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,0,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,0,0.009530666594703993
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,0,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,40,40,128,1,float16,fp8,0,0.013274667163689932
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,40,40,128,1,float16,float16,0,0.012256000190973282
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,0,0.011109333485364914
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,0,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,0,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,0,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,40,40,128,1,float16,float16,0,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,40,40,128,1,float16,fp8,0,0.012938667088747025
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,0,0.010650667051474253
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,0,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,0,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,0,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,0,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,40,40,128,1,float16,float16,0,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,40,40,128,1,float16,fp8,0,0.012629333883523941
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,0,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,0,0.009455999980370203
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,0,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,0,0.009658666948477427
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,0,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,0,0.01552533358335495
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,0,0.015002666662136713
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,0,0.018138666947682697
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,0,0.017466666797796886
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,0,0.024512000381946564
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,32,128,1,float16,float16,0,0.05606933434804281
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,32,128,1,float16,fp8,0,0.039647998909155525
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,0,0.028912000358104706
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,0,0.012986666212479273
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,0,0.012730666746695837
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,0,0.01309866706530253
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,0,0.018709332992633183
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,0,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,0,0.013343999783198038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,32,128,1,float16,fp8,0,0.025279998779296875
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,0,0.01116266722480456
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,0,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,0,0.011173332730929056
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,0,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,0,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,0,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,32,128,1,float16,fp8,0,0.017301333447297413
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,32,128,1,float16,float16,0,0.019920000185569126
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,0,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,0,0.00955200009047985
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,0,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,0,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,0,0.011450666934251785
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,0,0.01129066695769628
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,32,128,1,float16,float16,0,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,32,128,1,float16,fp8,0,0.014949332922697067
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,0,0.010181333248813948
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,0,0.010618666807810465
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,0,0.009413333609700203
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,0,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,0,0.01091733326514562
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,0,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,32,32,128,1,float16,float16,0,0.027349332968393963
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,32,128,1,float16,float16,0,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,32,128,1,float16,fp8,0,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,0,0.009242666885256767
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,0,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,0,0.009429333110650381
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,0,0.009882666791478792
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,0,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,32,128,1,float16,fp8,0,0.011301333705584208
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,32,128,1,float16,float16,0,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,0,0.010421333213647207
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,0,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,0,0.009759999811649323
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,0,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,0,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,0,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,32,128,1,float16,fp8,0,0.011183999478816986
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,32,128,1,float16,float16,0,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,0,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,0,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,0,0.009914666414260864
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,0,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,0,0.010053333515922228
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,0,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,32,128,1,float16,fp8,0,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,32,128,1,float16,float16,0,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,0,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,0,0.01080000028014183
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,0,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,0,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,0,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,0,0.010512000570694605
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,0,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,0,0.015013333410024643
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,0,0.017445333302021027
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,0,0.01907733331123988
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,0,0.015354666858911514
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,24,128,1,float16,float16,0,0.04271466533342997
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,0,0.02755733331044515
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,24,128,1,float16,fp8,0,0.033488000432650246
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,0,0.023743999501069386
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,0,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,0,0.0120319997270902
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,0,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,0,0.017322666943073273
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,24,128,1,float16,float16,0,0.02327466756105423
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,0,0.01732800031701724
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,0,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,0,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,0,0.01118933285276095
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,0,0.01137599969903628
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,0,0.01119999960064888
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,0,0.013642666240533194
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,0,0.013269333789745966
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,24,128,1,float16,float16,0,0.01714666684468587
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,0,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,0,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,24,24,128,1,float16,fp8,0,0.0222080002228419
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,0,0.011098666737476984
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,0,0.011551999797423681
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,0,0.011519999553759893
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,0,0.01119999960064888
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,24,128,1,float16,float16,0,0.01339200014869372
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,24,128,1,float16,fp8,0,0.013749333719412485
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,0,0.010234666367371878
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,0,0.011237333218256632
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,24,24,128,1,float16,fp8,0,0.016949333250522614
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,0,0.009466666728258133
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,0,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,0,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,24,128,1,float16,float16,0,0.011519999553759893
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,24,128,1,float16,fp8,0,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,0,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,0,0.009242666885256767
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,0,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,0,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,0,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,0,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,24,128,1,float16,float16,0,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,24,128,1,float16,fp8,0,0.011349332829316458
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,0,0.009173333023985228
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,0,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,0,0.009935999910036722
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,0,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,0,0.0106133334338665
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,0,0.009232000137368837
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,24,128,1,float16,float16,0,0.010309333602587381
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,0,0.010090666512648264
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,24,128,1,float16,fp8,0,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,0,0.00927466650803884
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,0,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,0,0.009290666629870733
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,0,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,0,0.009648000200589498
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,0,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,24,128,1,float16,fp8,0,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,24,128,1,float16,float16,0,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,0,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,0,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,0,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,0,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,0,0.010026666646202406
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,0,0.009248000259200731
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,0,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,0,0.013818666338920593
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,0,0.013258667041858038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,0,0.01350933313369751
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,0,0.019018666197856266
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,0,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,0,0.02717333287000656
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,16,128,1,float16,float16,0,0.02757866680622101
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,16,128,1,float16,fp8,0,0.02390933285156886
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,0,0.011488000551859537
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,0,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,0,0.025274666647116344
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,0,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,0,0.011898666620254517
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,0,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,0,0.013114667187134424
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,0,0.018426666657129925
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,16,128,1,float16,float16,0,0.01814933369557063
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,0,0.0170666662355264
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,16,128,1,float16,fp8,0,0.017029333859682083
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,0,0.010496000448862711
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,0,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,0,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,0,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,0,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,0,0.013349333157142004
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,0,0.013269333789745966
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,16,128,1,float16,float16,0,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,16,128,1,float16,fp8,0,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,0,0.01044800008336703
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,0,0.009642666826645533
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,0,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,0,0.01027199998497963
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,0,0.010138666878143946
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,0,0.009888000165422758
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,0,0.010597333312034607
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,0,0.011434666812419891
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,16,128,1,float16,fp8,0,0.011317333827416102
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,16,128,1,float16,float16,0,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,0,0.011370666325092316
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,0,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,0,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,0,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,0,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,0,0.010453333457310995
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,0,0.011152000476916632
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,0,0.010138666878143946
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,16,128,1,float16,float16,0,0.010703999549150467
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,0,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,0,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,0,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,0,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,0,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,0,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,0,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,0,0.010138666878143946
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,16,128,1,float16,float16,0,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,0,0.009322666873534521
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,16,128,1,float16,fp8,0,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,0,0.009237333511312803
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,0,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,16,16,128,1,float16,fp8,0,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,0,0.010154666379094124
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,0,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,0,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,0,0.010266666611035665
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,0,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,0,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,16,128,1,float16,fp8,0,0.010608000059922537
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,0,0.009477333476146063
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,16,128,1,float16,float16,0,0.010341333225369453
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,0,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,0,0.009445333232482275
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,0,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,0,0.009786666681369146
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,0,0.009557333464423815
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,0,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,0,0.009808000177145004
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,0,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,16,128,1,float16,float16,0,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,0,0.009749333063761393
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,16,128,1,float16,fp8,0,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,0,0.009392000113924345
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,0,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,0,0.009632000078757605
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,0,0.009461333354314169
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,0,0.01044800008336703
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,0,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,0,0.013242666920026144
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,0,0.012346666306257248
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,0,0.012778667112191519
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,0,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,12,128,1,float16,float16,0,0.023130667706330616
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,0,0.01802666609485944
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,0,0.01699200024207433
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,0,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,0,0.010938666760921478
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,0,0.013072000195582708
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,0,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,12,128,1,float16,fp8,0,0.02161066730817159
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,0,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,12,128,1,float16,float16,0,0.01653333380818367
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,0,0.012991999586423239
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,0,0.010640000303586325
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,12,128,1,float16,fp8,0,0.01573866605758667
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,0,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,0,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,0,0.011312000453472137
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,0,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,0,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,12,128,1,float16,float16,0,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,12,128,1,float16,fp8,0,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,0,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,0,0.00983466642598311
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,0,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,0,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,0,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,12,128,1,float16,float16,0,0.011968000481526056
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,12,128,1,float16,fp8,0,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,0,0.010245333115259806
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,0,0.009525333220760027
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,0,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,0,0.00980266680320104
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,0,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,0,0.010618666807810465
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,0,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,0,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,0,0.009429333110650381
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,0,0.009712000067035357
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,0,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,0,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,0,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,12,128,1,float16,float16,0,0.010469333579142889
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,12,128,1,float16,fp8,0,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,12,128,1,float16,float16,0,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,0,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,0,0.010133333504199982
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,0,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,12,12,128,1,float16,fp8,0,0.011413333316644033
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,0,0.009674666449427605
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,0,0.009375999992092451
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,12,128,1,float16,fp8,0,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,12,128,1,float16,float16,0,0.009365333244204521
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,0,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,0,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,0,0.00996800015370051
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,0,0.009189333145817121
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,0,0.011183999478816986
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,12,128,1,float16,float16,0,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,12,128,1,float16,fp8,0,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,0,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,0,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,0,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,0,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,0,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,0,0.009349333122372627
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,0,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,0,0.010037333394090334
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,0,0.011493333925803503
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,0,0.013183999806642532
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,0,0.013354666531085968
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,0,0.019029332945744198
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,0,0.019189332922299702
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,0,0.017024000485738117
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,0,0.017312000195185345
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,0,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,0,0.011370666325092316
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,0,0.01137599969903628
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,0,0.011322667201360067
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,0,0.01471466695268949
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,0,0.01301866645614306
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,0,0.012234666695197424
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,0,0.019306667149066925
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,0,0.017258666455745697
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,0,0.013647999614477158
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,0,0.01333333303531011
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,0,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,0,0.011109333485364914
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,0,0.0102613332370917
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,0,0.009813333551088968
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,0,0.011407999942700068
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,0,0.011274666835864386
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,0,0.01458666721979777
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,0,0.014111999422311783
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,0,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,0,0.009685333197315535
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,0,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,0,0.011114666859308878
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,0,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,0,0.01119999960064888
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,0,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,0,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,0,0.011413333316644033
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,0,0.009946666657924652
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,0,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,0,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,0,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,0,0.011215999722480774
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,0,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,0,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,0,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,0,0.009173333023985228
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,0,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,0,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,0,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,0,0.010506667196750641
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,0,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,0,0.0100426667680343
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,0,0.009285333255926767
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,0,0.009626666704813639
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,0,0.0106133334338665
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,0,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,0,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,0,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,0,0.009242666885256767
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,0,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,0,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,0,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,0,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,0,0.009519999846816063
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,0,0.009279999881982803
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,0,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,0,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,0,0.010064000263810158
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,0,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,0,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,0,0.010090666512648264
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,0,0.009232000137368837
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,0,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,0,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,0,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,0,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,0,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,0,0.011328000575304031
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,0,0.00921066664159298
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,0,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,0,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,0,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,0,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,0,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,0,0.010186666622757912
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,0,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,0,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,0,0.014858666807413101
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,0,0.013429333766301474
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,0,0.013338666409254074
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,0,0.013562666873137156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,0,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,0,0.011312000453472137
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,0,0.011146667102972666
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,0,0.011237333218256632
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,0,0.015168000012636185
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,0,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,0,0.012186666329701742
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,0,0.011359999577204386
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,0,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,0,0.010266666611035665
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,0,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,0,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,0,0.009786666681369146
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,0,0.011221333096424738
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,0,0.011445333560307821
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,0,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,0,0.010255999863147736
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,0,0.009973333527644476
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,0,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,0,0.00927466650803884
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,0,0.010170666500926018
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,0,0.011237333218256632
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,0,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,0,0.011114666859308878
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,0,0.010106666634480158
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,0,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,0,0.010015999898314476
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,0,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,0,0.009642666826645533
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,0,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,0,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,0,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,0,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,0,0.009882666791478792
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,0,0.009904000287254652
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,0,0.010138666878143946
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,0,0.009872000043590864
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,0,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,0,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,0,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,0,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,0,0.010149333626031876
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,0,0.010250666489203772
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,0,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,0,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,0,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,0,0.010309333602587381
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,0,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,0,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,0,0.009999999776482582
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,0,0.010368000095089277
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,0,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,0,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,0,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,0,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,0,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,0,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,0,0.00921066664159298
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,0,0.00921066664159298
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,0,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,0,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,0,0.011338666081428528
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,0,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,0,0.01201066623131434
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,0,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,0,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,0,0.011541333049535751
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,0,0.011317333827416102
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,0,0.009808000177145004
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,0,0.011173332730929056
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,0,0.009290666629870733
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,0,0.011546666423479715
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,0,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,0,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,0,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,0,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,0,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,0,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,0,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,0,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,0,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,0,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,0,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,0,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,0,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,0,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,0,0.009375999992092451
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,0,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,0,0.01007466639081637
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,0,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,0,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,0,0.01002133327225844
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,0,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,0,0.010122666756312052
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,0,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,0,0.010266666611035665
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,0,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,0,0.00980266680320104
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,0,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,0,0.009402666861812273
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,0,0.009839999799927076
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,0,0.009658666948477427
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,0,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,0,0.009402666861812273
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,0,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,0,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,0,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,0,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,0,0.009349333122372627
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,0,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,0,0.013199999928474426
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,0,0.011312000453472137
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,0,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,0,0.011183999478816986
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,0,0.009663999701539675
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,0,0.010485333700974783
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,0,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,0,0.011930666863918304
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,0,0.009717333440979322
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,0,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,0,0.009679999823371569
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,0,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,0,0.009573333586255709
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,0,0.009375999992092451
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,0,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,0,0.009509333098928133
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,0,0.016869333883126576
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,0,0.009583999713261923
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,0,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,0,0.009632000078757605
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,0,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,0,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,0,0.009530666594703993
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,0,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,0,0.009248000259200731
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,0,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,0,0.009610666582981745
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,0,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,0,0.009818666925032934
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,0,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA GB300,context_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,0,0.009162666896979014
