🐎

Performance Benchmark

Public

main

17m

[V1][Spec Decode] Remove deprecated spec decode config params (#15466)#11601

shangmingc·main 239b7bef·Created today at 4:19 PM

35m

[Bugfix] Check dimensions of multimodal embeddings in V1 (#15816)#11600

Cyrus Leung·main 09e974d4·Created today at 4:01 PM

37m

Upgrade `transformers` to `v4.50.3` (#13905)#11599

Harry Mellor·main e5ef4fa9·Created today at 3:59 PM

[Bugfix] Fix missing return value in load_weights method of adapters.py (#15542)#11598

Mrm·main 037bcd94·Created today at 1:56 PM

[Bugfix] Fix Crashing When Loading Modules With Batchnorm Stats (#15813)#11597

Alex Brooks·main c2e7507a·Created today at 1:23 PM

neuralmagic:lwilkinson/no-pad-fa3

update fa#11473

Lucas Wilkinson·neuralmagic:lwilkinson/no-pad-fa3 ab4a53cc·Created Tuesday at 11:09 PM

add scheduler code#11424

Lucas Wilkinson·neuralmagic:lwilkinson/no-pad-fa3 e1ead1c9·Created Mon 24th Mar at 4:34 AM

Merge remote-tracking branch 'origin/main' into lwilkinson/no-pad-fa3#11423

Lucas Wilkinson·neuralmagic:lwilkinson/no-pad-fa3 6285e78e·Created Mon 24th Mar at 4:24 AM

update mla#11398

Lucas Wilkinson·neuralmagic:lwilkinson/no-pad-fa3 705ea02a·Created Sat 22nd Mar at 4:59 AM

more efficient v1 pad switch#11060

Lucas Wilkinson·neuralmagic:lwilkinson/no-pad-fa3 a0262e00·Created Fri 7th Mar at 2:49 AM

Shaoting-Feng:v1-benchmark

trigger buildkite#10722

Shaoting·Shaoting-Feng:v1-benchmark e6ae6ec2·Created Tue 18th Feb at 6:30 PM

Merge branch 'main' into v1-benchmark#10568

Kuntai Du·Shaoting-Feng:v1-benchmark 7852dd53·Created Tue 11th Feb at 2:44 AM

Merge branch 'main' into v1-benchmark#10533

Kuntai Du·Shaoting-Feng:v1-benchmark 7852dd53·Created Sat 8th Feb at 5:49 AM

simon-mo:h200-v1-bench

use premerge image#10232

Simon Mo·simon-mo:h200-v1-bench 0dbe0623·Created Fri 17th Jan at 11:59 PM

31m

support pre-merge workflow#10231

Simon Mo·simon-mo:h200-v1-bench 51255850·Created Fri 17th Jan at 10:52 PM

trigger new commit#10214

Simon Mo·simon-mo:h200-v1-bench 49452750·Created Thu 16th Jan at 7:29 PM

Merge branch 'main' of github.com:vllm-project/vllm into h200-v1-bench#10208

Simon Mo·simon-mo:h200-v1-bench 24d2dfe4·Created Thu 16th Jan at 10:13 AM

actually, run on all hardwares#9565

Simon Mo·simon-mo:h200-v1-bench 5c6e2c73·Created Fri 22nd Nov 2024 at 7:42 AM

jikunshang:add_genai_perf

10h

update dependency#10108

Kunshang Ji·jikunshang:add_genai_perf 5687ffdd·Created Thu 9th Jan at 12:50 AM

—

[CI]add genai-perf benchmark in nightly benchmark#10107

Kunshang Ji·jikunshang:add_genai_perf 816143a8·Created Thu 9th Jan at 12:50 AM

format#10080

Kunshang Ji·jikunshang:add_genai_perf f89fcebf·Created Tue 7th Jan at 11:53 PM

13s

format#10079

Kunshang Ji·jikunshang:add_genai_perf ee2db6ad·Created Tue 7th Jan at 11:39 PM

format#10022

Kunshang Ji·jikunshang:add_genai_perf 07666612·Created Fri 3rd Jan at 1:31 AM

simon-mo:h100-bench-v2

newline#9558

Simon Mo·simon-mo:h100-bench-v2 75f77c85·Created Fri 22nd Nov 2024 at 1:55 AM

restor skip#9557

Simon Mo·simon-mo:h100-bench-v2 b7e8979a·Created Fri 22nd Nov 2024 at 1:54 AM

Fix reporting, and try again#9555

Simon Mo·simon-mo:h100-bench-v2 6490d4bd·Created Thu 21st Nov 2024 at 10:31 PM

54m

Update benchmark pipeline for H100#9549

Simon Mo·simon-mo:h100-bench-v2 2cdd865d·Created Thu 21st Nov 2024 at 7:30 PM

simon-mo:h200-bench

11h

restore other changes#9523

Simon Mo·simon-mo:h200-bench d88811a8·Created Wed 20th Nov 2024 at 6:33 AM

48m

rerun and sort tables#9515

Simon Mo·simon-mo:h200-bench d70bc024·Created Wed 20th Nov 2024 at 4:19 AM

cache models#9513

Simon Mo·simon-mo:h200-bench 42928a7f·Created Wed 20th Nov 2024 at 1:37 AM

carry on#9510

Simon Mo·simon-mo:h200-bench 474a3cef·Created Wed 20th Nov 2024 at 12:30 AM

29m

kuntai's suggesitons#9509

Simon Mo·simon-mo:h200-bench 6923dd4f·Created Tue 19th Nov 2024 at 11:26 PM

KuntaiDu:kuntai-update-nightlybench

Bump up version of all containers#8880

Kuntai Du·KuntaiDu:kuntai-update-nightlybench 4f1a72a1·Created Fri 4th Oct 2024 at 4:24 AM

Merge branch 'kuntai-update-nightlybench' of https://github.com/KuntaiDu/vllm into kuntai-update-nightlybench#8873

Kuntai Du·KuntaiDu:kuntai-update-nightlybench 141ce129·Created Thu 3rd Oct 2024 at 9:27 PM

bug fix: latency has been removed.#8870

Kuntai Du·KuntaiDu:kuntai-update-nightlybench e20d6b98·Created Thu 3rd Oct 2024 at 5:48 PM

update metric collection metric to incorporate latest benchmark_serving.py#8863

Kuntai Du·KuntaiDu:kuntai-update-nightlybench 9c20da08·Created Wed 2nd Oct 2024 at 10:10 PM

Merge branch 'kuntai-update-nightlybench' of https://github.com/KuntaiDu/vllm into kuntai-update-nightlybench#8818

Kuntai Du·KuntaiDu:kuntai-update-nightlybench 5bc01989·Created Fri 27th Sep 2024 at 8:32 PM

simon-mo:timeout-wait-for-images

13h

yaml#8664

Simon Mo·simon-mo:timeout-wait-for-images 52311f8e·Created Mon 16th Sep 2024 at 8:11 PM

43m

[perf bench] set timeout to debug hanging#8663

Simon Mo·simon-mo:timeout-wait-for-images 2d4cbddb·Created Mon 16th Sep 2024 at 7:27 PM

KuntaiDu:kuntai-cleanup-perfbench

adjust the command to kill vllm#8272

Kuntai Du·KuntaiDu:kuntai-cleanup-perfbench 548cf408·Created Mon 19th Aug 2024 at 5:04 PM

adjust the command to kill vllm#8269

Kuntai Du·KuntaiDu:kuntai-cleanup-perfbench 19f539e5·Created Mon 19th Aug 2024 at 4:21 AM

bug fix#8266

Kuntai Du·KuntaiDu:kuntai-cleanup-perfbench 8930c9ee·Created Mon 19th Aug 2024 at 12:34 AM

51m

kill process related to port 8000#8263

Kuntai Du·KuntaiDu:kuntai-cleanup-perfbench 69e8f6a7·Created Sun 18th Aug 2024 at 11:42 PM

48m

adjust how to kill the processes#8262

Kuntai Du·KuntaiDu:kuntai-cleanup-perfbench 93e57911·Created Sun 18th Aug 2024 at 10:50 PM