55 Commits
Author SHA1 Message Date
31e90ac370 [Attention] Enable FA4 sparse MQA decode on Blackwell (#55866)
Signed-off-by: Simon Veitner <sveitner@redhat.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
2026-10-13 09:31:37 -04:00
91d7324cb1 [perf] wire FA and FlashMLA for sm90 GLM5Next NoPE SparseMLA (#55385)
Signed-off-by: JaredforReal <w13431838023@gmail.com>
Co-authored-by: Kimi Code <noreply@moonshot.cn>
Co-authored-by: Leoyzen <leoyzen@gmail.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
2026-09-22 02:03:59 -07:00
GroupieSteven 4ee2595512 [Attention] Sync FA with upstream (#54819)
Signed-off-by: Chuyue Wang <stevenwang0805@outlook.com>
2026-09-04 23:48:32 -04:00
Matthew BonanniandOpenAI Codex de9250ac9e [Attention] Enable masked MHA for GLM-5 head dimensions (#53785)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
2026-08-27 21:17:12 +00:00
simon-veitner-redhat 7156c63bef [SM100] Hdim 256 optimized (#52980)
Signed-off-by: Simon Veitner <sveitner@redhat.com>
2026-08-25 20:28:40 -07:00
Roger Wang 755492e37d Revert "[Kernel] Gemma-4 FA4 FP8 Kernel" (#52987) 2026-08-19 10:48:46 -07:00
827a2af806 [Kernel] Gemma-4 FA4 FP8 Kernel (#48666)
Signed-off-by: Qixiang Lin <qixiangl@nvidia.com>
Signed-off-by: Jhao-Ting Chen <jhaotingc@nvidia.com>
Co-authored-by: Qixiang Lin <qixiangl@nvidia.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Yongye Zhu <zyy1102000@gmail.com>
2026-08-13 17:42:00 -07:00
Chris LeonardandShengqi Chen 41e7746b82 Update vllm to point to flash-attention commit that builds FA3 with torch stable API. (Retry) (#49599)
Signed-off-by: Chris Leonard <chleonar@redhat.com>
Signed-off-by: Shengqi Chen <harry-chen@outlook.com>
Co-authored-by: Shengqi Chen <harry-chen@outlook.com>
2026-08-06 08:37:46 -07:00
Jane (Yuan) Xu eb6453d95b [Build] Update pin to build ABI stable FA2 (#50474) 2026-07-31 20:09:36 -07:00
Matthew BonanniandOpenAI Codex 82ae4164ee [2/N][Attention] Enable masked MHA for sparse MLA prefills (#48770)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
2026-07-31 06:01:24 -07:00
Andrey Talman c79ff5f918 [Build] Bump vllm-flash-attn to C++20-compatible commit for torch-nightly (#49326)
Signed-off-by: Andrey Talman <atalman@fb.com>
2026-07-22 13:51:59 +00:00
Matthew Bonanniandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> bd091079cb [Attention] FlashAttention 4 SM100 FP8 kv cache support (#42569)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-07-20 10:53:27 -07:00
Artem PerevedentsevandMatthew Bonanni 7cd1d57b74 [CI/Build][Docker] Bump nvidia-cutlass-dsl to 4.6.0 and drop packaging workarounds (#47442)
Signed-off-by: Artem Perevedentsev <aperevedents@nvidia.com>
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
2026-07-16 13:51:41 +00:00
Lucas Wilkinson bec0a4ede6 [Revert] [Build] Update vllm ...builds FA3 with torch stable API (#48269)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2026-07-11 05:20:25 +00:00
Chris Leonard 8347c6e6e1 updated flash_attn GIT_TAG to point to torch Stable ABI FA3 commit (#47995)
Signed-off-by: Chris Leonard <chleonar@redhat.com>
2026-07-08 10:56:29 -07:00
Roberto L. CastroandLucas Wilkinson 248d1fbb71 [Feat][1/N] CuTeDSL warmup infrastructure, FA4 MLA (#46182)
Signed-off-by: LopezCastroRoberto <rocastro@redhat.com>
Signed-off-by: Roberto L. Castro <38211239+LopezCastroRoberto@users.noreply.github.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
2026-06-30 12:17:34 -07:00
Chris Leonard d0f800811b [Build] Update vllm to point to vllm-project/flash-attention commit that builds FA3 with torch stable API. (#46644) 2026-06-26 19:42:46 -07:00
eb28452b10 [Model] Add DiffusionGemma Support (#45163)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Martin Kukla <martin.kukla@cantab.net>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Dipika Sikka <dsikka@redhat.com>
Co-authored-by: NickLucche <nlucches@redhat.com>
Co-authored-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
Co-authored-by: Alec Kohlhoff <134344302+aleckohlhoff@users.noreply.github.com>
Co-authored-by: Porras Huang <20535584+porrashuang@users.noreply.github.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: scoootscooob <167050519+scoootscooob@users.noreply.github.com>
2026-06-11 22:17:35 -07:00
Matthew Bonanni ea0d045a05 [FlashAttention] Sync FA with upstream (#44065)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
2026-06-02 07:15:37 -07:00
Matthew Bonanni dcacdf9a88 [Attention] Sync FA with upstream (#41052) 2026-05-12 23:34:18 -04:00
LainandLucas Wilkinson e24e0a43a4 [Attention] relax the head dim 512 and paged kv for sm90+FA4 (#38835)
Signed-off-by: Siyuan Fu <siyuanf@nvidia.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
2026-04-08 18:23:18 +00:00
Matthew Bonanni 308cec5864 [FlashAttention] Symlink FA4 instead of copying when using VLLM_FLASH_ATTN_SRC_DIR (#38814)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
2026-04-08 12:04:34 +00:00
Lucas Wilkinson cb3935a8fc [FA4] Update flash-attention to latest upstream FA4 (#38690)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2026-04-02 17:02:37 +00:00
Jonas M. Kübler 77d2a5f17b pick up tuned prefill configs for FP8 FA3 (#36265)
Signed-off-by: Jonas M. Kübler <44084297+jmkuebler@users.noreply.github.com>
Signed-off-by: Jonas Kuebler <kuebj@amazon.com>
2026-03-17 07:00:26 -07:00
Matthew Bonanni f444c05c32 [Attention] Use FA4 for MLA prefill (#34732)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
2026-03-12 12:10:17 -04:00
Lucas WilkinsonandWentao Ye f44d1ddc8c [BugFix] Fix cmake based incremental install (wrong vllm install dir) (#35773)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
2026-03-02 21:58:16 -08:00
8b5014d3dd [Attention] FA4 integration (#32974)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
2026-03-01 23:44:57 +00:00
Lucas Wilkinson c7914d30f9 Reapply [Attention][FA3] Update FA3 to include new swizzle optimization (#34043)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2026-02-11 07:07:56 -08:00
Luka Govedič e3bf79ffa0 Revert "[Attention][FA3] Update FA3 to include new swizzle optimization" (#33841) 2026-02-04 19:54:27 -08:00
Lucas Wilkinson 2267cb1cfd [Attention][FA3] Update FA3 to include new swizzle optimization (#23465)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2026-02-03 08:08:47 -08:00
Lucas Wilkinson be6a81f31b [chore] Update FA commit (#30460)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2026-01-07 23:24:18 -08:00
Lucas Wilkinson c68c7b403d [BugFix] Fix missing symbol triggering FA2 fallback on Hopper (#29107)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2025-11-21 13:58:32 -08:00
Matthew Bonanni 4c23690f43 [Attention] FlashAttention ViT support, make default backend (#28763)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
2025-11-18 20:06:21 -08:00
Matthew BonanniandCyrus Leung 8cc40f8992 [Attention] Bump FA for removed method (#28429)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
2025-11-14 09:13:37 -08:00
Jonas M. KüblerandLucas Wilkinson 9c84ca8293 [FA/Chore] Bump FA version for FP8 two-level accumulation (#27889)
Signed-off-by: Jonas Kuebler <kuebj@amazon.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
2025-11-10 12:06:04 -08:00
Lucas Wilkinson 9f020f4f31 [BugFix] Fix failing gemma-3-1b-it test: test_lm_eval_accuracy_v1_engine[google/gemma-3-1b-it] (#27111)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2025-10-18 12:44:39 -06:00
Ming Yang 3b736e1c38 [Attention][DCP] Support DCP with query length > 1 (MTP) with FA3 (#25049)
Signed-off-by: Ming Yang <minos.future@gmail.com>
2025-10-09 08:06:29 -07:00
Lucas Wilkinson 418d111f8c [FA/Chore] Bump vllm-flash-attention (#25537)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2025-10-02 11:06:14 -04:00
402759d472 [Attention] FlashAttn MLA (#14258)
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Matthew Bonanni <mbonanni001@gmail.com>
Co-authored-by: Matthew Bonanni <mbonanni001@gmail.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
2025-09-04 02:47:59 -07:00
Lucas Wilkinson 292084e72a [BugFix] Fix for IMA in FA3 varlen combine (#22967)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2025-08-17 08:52:04 -07:00
Lucas Wilkinson 177e55e3bd [Attention] FA3 Attention Sinks Perf Boost (#22478)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
2025-08-15 17:41:07 -04:00
Thomas Parnell bd875d2eb7 [Bugfix] Update FA commit hash (#22546)
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
2025-08-08 16:10:25 -07:00
Lucas Wilkinson 2cb6ef8996 [BugFix] Fix FA2 RuntimeError when sinks is provided (#22365)
Signed-off-by: LucasWilkinson <lwilkinson@neuralmagic.com>
2025-08-06 08:03:03 -07:00
Woosuk Kwon e3c876dca3 Upgrade FA3 for attention sink (#22313)
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
2025-08-05 21:36:21 -07:00
Woosuk Kwon 8acb4badee [CUDA graphs] Enable full cuda graphs with FA3 AoT scheduling (#20301)
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
2025-07-01 09:07:36 -07:00
Eli Uriegas 0d06b533a0 cmake: Update vllm_flash_attn for vllm_kernels (#20032)
Signed-off-by: Eli Uriegas <eliuriegas@meta.com>
2025-06-24 22:44:10 +00:00
Lucas Wilkinson a045b7e89a [Perf] Improve/Fix-regression for FA3 in High QPS regimes (#19463)
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
2025-06-24 13:09:01 -04:00
Lucas Wilkinson 07334959d8 [Wheel Size] Only build FA2 8.0+PTX (#19336) 2025-06-17 12:32:49 +09:00
Luka Govedič a3896c7f02 [Build] Fixes for CMake install (#18570) 2025-05-27 20:49:24 -04:00
yexin(叶鑫)andMagnetoWang b22980a1dc [Perf]Optimize rotary_emb implementation to use Triton operator for improved inference performance (#16457)
Signed-off-by: cynthieye <yexin93@qq.com>
Co-authored-by: MagnetoWang <magnetowang@outlook.com>
2025-04-25 14:52:28 +08:00