Commits · 2108a571d7ee522346d309ed902e9da8e7cdc97b · OpenDAS / vllm_cscc

08 Nov, 2025 2 commits

[DCP] Support dcp kv_cache interleave size > 1 (#26696) · 2108a571

zhangsicheng5 authored Nov 09, 2025


Signed-off-by: zhangsicheng5 <zhangsicheng5@huawei.com>
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com>
Signed-off-by: Qiu <qiuchunshuo@huawei.com>
Co-authored-by: QiuChunshuo <qiuchunshuo@huawei.com>

2108a571

[Attention] Remove max cudagraph size limit of 992 (#27840) · 608bb144
22quinn authored Nov 07, 2025
```
Signed-off-by: 22quinn <33176974+22quinn@users.noreply.github.com>
```
608bb144

05 Nov, 2025 5 commits
- [FlashInfer] Avoid FlashInfer block_size 16 + head_size 256 on blackwell (#27994) · c765f0b4
  Chen Zhang authored Nov 05, 2025
```
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
```
  c765f0b4
- [ROCm][MLA] Support block-size > 1 for AITER MLA backend (#27224) · 6cae1e53
  Pleaplusone authored Nov 05, 2025
```
Signed-off-by: ganyi <ygan@amd.com>
Co-authored-by: wuhuikx <hattie.wu@amd.com>
```
  6cae1e53
- [bugfix] fix wrong `dcp_local_seq_lens` calc (#27518) · 16b37f31
  Qiu authored Nov 05, 2025
```
Signed-off-by: Qiu <qiuchunshuo@huawei.com>
```
  16b37f31
- [BugFix] Fix DCP Assert (AssertionError: DCP not support reorder_batch_threshold > 1 now.) (#28100) · d43ad5a7
  Lucas Wilkinson authored Nov 05, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  d43ad5a7
- [XPU] Add gpt-oss model support for Intel GPU (#27786) · 18b39828
  Kunshang Ji authored Nov 05, 2025
```
Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
```
  18b39828
04 Nov, 2025 2 commits
- [ROCm][Perf] New design on ROCm AITER MHA backend Implementation (#25763) · dc937175
  Pleaplusone authored Nov 05, 2025
```
Signed-off-by: ganyi <ygan@amd.com>
```
  dc937175
- [Bug] Batch invariant: Fix flash attn MLA `RuntimeError: scheduler_metadata... · 7e4be741
  Wentao Ye authored Nov 04, 2025
```
[Bug] Batch invariant: Fix flash attn MLA `RuntimeError: scheduler_metadata must have shape (metadata_size)` (#27884)
```
  7e4be741
03 Nov, 2025 2 commits
- [Bugfix] change FlashMLA reorder_batch_threshold (#27777) · 145c00a4
  Matthew Bonanni authored Nov 03, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  145c00a4
- [Hybrid] Pass kernel block size to builders (#27753) · 18961c5e
  Thomas Parnell authored Nov 03, 2025
```
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
```
  18961c5e
02 Nov, 2025 1 commit
- [V1] [Hybrid] Mamba1 Automatic Prefix Caching (#26377) · 00b31a36
  Asaf Joseph Gardin authored Nov 02, 2025
```
Signed-off-by: asafg <39553475+Josephasafg@users.noreply.github.com>
```
  00b31a36
31 Oct, 2025 2 commits
- Add FLASHINFER_MLA to test_mla_backends and add B200 CI run (#27663) · f29aeb5a
  Matthew Bonanni authored Oct 31, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  f29aeb5a
- [Bugfix] Avoid too small block m/n for FlexAttention kernel option (#27853) · 3933f18a
  Isotr0py authored Oct 31, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  3933f18a
30 Oct, 2025 1 commit
- [BugFix] Reordering extend logic fix (#27739) · b5d70751
  Lucas Wilkinson authored Oct 30, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  b5d70751
28 Oct, 2025 2 commits

[Misc] Make reorder batch also separate extends (#27367) · 141e6a05
Lucas Wilkinson authored Oct 29, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
141e6a05

[Bugfix][CI] Fix v1 attention backend tests and add CI coverage (#26597) · a8c02fb5

Mohammad Miadh Angkad authored Oct 28, 2025


Signed-off-by: Mohammad Miadh Angkad <MAngkad.BSDSBA2027@aim.edu>
Signed-off-by: Mohammad Miadh Angkad <mangkad.bsdsba2027@aim.edu>
Co-authored-by: Ye (Charlotte) Qi <yeq@meta.com>

a8c02fb5

26 Oct, 2025 1 commit

[Chore]:Extract math and argparse utilities to separate modules (#27188) · 71b1c8b6

Yeshwanth N authored Oct 26, 2025


Signed-off-by: Yeshwanth Surya <yeshsurya@gmail.com>
Signed-off-by: Yeshwanth N <yeshsurya@gmail.com>
Signed-off-by: yeshsurya <yeshsurya@gmail.com>

71b1c8b6

24 Oct, 2025 2 commits

[Attention] Add MLA prefill backend: trtllm_ragged_attention_deepseek (#26397) · 0f67d4d9
Ming Yang authored Oct 24, 2025
```
Signed-off-by: Ming Yang <minos.future@gmail.com>
```
0f67d4d9

[MISC] `cudagraph_capture_sizes` related improvements (#26016) · 284cc922

fhl2000 authored Oct 24, 2025

Signed-off-by: fhl <2410591650@qq.com>
Signed-off-by: fhl2000 <63384265+fhl2000@users.noreply.github.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

284cc922

23 Oct, 2025 2 commits
- [Attention] Fix FlashMLA metadata builder arguments for q_len > 1 (#27368) · dbfbf9f3
  Matthew Bonanni authored Oct 23, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  dbfbf9f3
- [Chore] Separate out `vllm.utils.platform_utils.py` (#27374) · ca76486a
  Jonathan Chen authored Oct 23, 2025
```
Signed-off-by: Jonathan <chenleejonathan@gmail.com>
```
  ca76486a
22 Oct, 2025 2 commits
- [BugFix] bugfix for Flash Attention MLA with full cuda graph IMA following pr-25490 (#27128) · 5beacce2
  Daisy-Ma-coder authored Oct 22, 2025
```
Signed-off-by: qqma <qqma@amazon.com>
Co-authored-by: qqma <qqma@amazon.com>
```
  5beacce2
- [Bugfix] Disable FlexAttention direct block mask building for encoder-only models (#27344) · 084a9dae
  Isotr0py authored Oct 23, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  084a9dae
21 Oct, 2025 1 commit

[Bugfix] fixes the decoding metadata of dense mla's fp8 kvcache. (#27144) · 250fb1b8

Tao He authored Oct 22, 2025


Signed-off-by: Tao He <linzhu.ht@alibaba-inc.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>

250fb1b8

18 Oct, 2025 2 commits
- [Chore] Clean up pytorch helper functions in `vllm.utils` (#26908) · 6ac5e06f
  Isotr0py authored Oct 19, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Signed-off-by: isotr0py <2037008807@qq.com>
```
  6ac5e06f
- [fix][cpu] fix prefill attention in CPU attention backend (#27035) · ab4be40f
  Fadi Arafeh authored Oct 18, 2025
```
Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com>
```
  ab4be40f
16 Oct, 2025 4 commits
- [small][batch invariance] Rename the env and internal flags to simplify usage (#26855) · b2f78cba
  Bram Wasti authored Oct 16, 2025
```
Signed-off-by: Bram Wasti <bwasti@meta.com>
```
  b2f78cba
- [Feature] Add process_weights_after_loading to AttentionImpl (#26870) · 5afd3276
  rongfu.leng authored Oct 16, 2025
```
Signed-off-by: rongfu.leng <rongfu.leng@daocloud.io>
```
  5afd3276
- Deepseek-v3 Batch Invariant on 8xH100 (#26609) · 7d8975de
  Bram Wasti authored Oct 15, 2025
```
Signed-off-by: Bram Wasti <bwasti@meta.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
```
  7d8975de
- [PERF] Qwen3-next MTP speedup (change bool mask indexing to index_select /... · 785d8b64
  Vadim Gimpelson authored Oct 16, 2025
```
[PERF] Qwen3-next MTP speedup (change bool mask indexing to index_select / index_copy to reduce d2h) (#26437)
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>
```
  785d8b64
15 Oct, 2025 1 commit

Move query quantization to attention layer for Flashinfer & Triton. (#26534) · 0a9ef0cf

Adrian Abeyta authored Oct 15, 2025


Signed-off-by: adabeyta <aabeyta@redhat.com>
Signed-off-by: Adrian Abeyta <aabeyta@redhat.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>

0a9ef0cf

14 Oct, 2025 3 commits

remove attn output view kernel (#26680) · a86b4c58

Boyuan Feng authored Oct 14, 2025

Signed-off-by: Boyuan Feng <boyuan@meta.com>
Signed-off-by: Boyuan Feng <fby.1994@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

a86b4c58

[Attention][Spec Decode] FlashMLA spec decode support (#26541) · 82af928c
Matthew Bonanni authored Oct 14, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
82af928c

[DCP] Support Decode Context Parallel (DCP) for GQA with FlashAttention (#24864) · ea97940d

Jaya Yuan authored Oct 14, 2025


Signed-off-by: yuanyongjie.yyj <yuanyongjie.yyj@antgroup.com>
Signed-off-by: FENP <32334296+FENP@users.noreply.github.com>
Signed-off-by: Jaya Yuan <yuanyongjie.yyj@antgroup.com>

ea97940d

13 Oct, 2025 2 commits
- [Log] Optimize Startup Log (#26601) · e251e457
  Wentao Ye authored Oct 13, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  e251e457
- [unrevert] Add batch invariant kernel override for FlashInfer backend [2/n] (#26373) · 32637990
  Bram Wasti authored Oct 13, 2025
```
Signed-off-by: Bram Wasti <bwasti@meta.com>
Signed-off-by: Bram Wasti <bwasti@fb.com>
```
  32637990
12 Oct, 2025 1 commit
- Update `Optional[x]` -> `x | None` and `Union[x, y]` to `x | y` (#26633) · 8fcaaf6a
  Harry Mellor authored Oct 12, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  8fcaaf6a
10 Oct, 2025 1 commit
- [deepseek] kernel block size for UniformTypeKVCacheSpecs (#26559) · 6f0f570c
  Chen Zhang authored Oct 10, 2025
```
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
```
  6f0f570c
09 Oct, 2025 1 commit
- [Bugfix] Fix CUDA graph selection bug in FlashInfer at high concurrency (#26499) · 6e783bc5
  Benjamin Chislett authored Oct 09, 2025
```
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
```
  6e783bc5