Commits · 6f1355a1b74e4502e6a4e6ba9a811cc50729ee1f · OpenDAS / vllm_cscc

25 Nov, 2025 1 commit
- [Perf] Disable DeepGEMM MoE by default when TP=8 is used (#29346) · 6f1355a1
  Michael Goin authored Nov 24, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  6f1355a1
24 Nov, 2025 1 commit
- [MoE][Refactor] Make select_experts a non-static method (#29067) · 8f066146
  bnellnm authored Nov 24, 2025
```
Signed-off-by: Bill Nell <bnell@redhat.com>
```
  8f066146
21 Nov, 2025 1 commit

[Log] Optimize startup log (#28948) · a42ab317

Wentao Ye authored Nov 21, 2025


Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Co-authored-by: Nick Hill <nhill@redhat.com>

a42ab317

19 Nov, 2025 1 commit

[Feat][Perf] Enable deepep-low-latency with round-robin expert placement. (#28449) · da2f6800

Chen Bruce authored Nov 19, 2025


Signed-off-by: bruceszchen <bruceszchen@tencent.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

da2f6800

14 Nov, 2025 1 commit

[Kernels] Enable FlashInfer FP8 Blockscale on SM90 (for TEP DSR1) (#27134) · 3f8a8740

Duncan Moss authored Nov 14, 2025

Signed-off-by: Duncan Moss <djm.moss@gmail.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>

3f8a8740

12 Nov, 2025 1 commit

[Performance][B200] Fix deepgemm prologue (#27897) · 74a9a9fa

Varun Sundar Rabindranath authored Nov 12, 2025


Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>

74a9a9fa

11 Nov, 2025 3 commits
- Remove weight_scale.T special case for SM90 Block FP8 CUTLASS kernel (#28431) · f9a40871
  Michael Goin authored Nov 11, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  f9a40871
- [Feature] Add env var `VLLM_MOE_USE_DEEP_GEMM` (#28422) · de540c03
  Wentao Ye authored Nov 10, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  de540c03
- [Feature] Refactor batch invariant fp8 DeepGEMM (#27606) · 35d801f1
  Wentao Ye authored Nov 10, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  35d801f1
10 Nov, 2025 2 commits

[Performance] Support FP8 flashinfer TRTLLM MOE on Qwen3 and Qwen-3next (#27492) · 34553b9d
jiahanc authored Nov 10, 2025
```
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
```
34553b9d

[RFC][ROCm][AITER] Keep all AITER kernels in `_aiter_ops` class like... · f080a835

vllmellm authored Nov 10, 2025


[RFC][ROCm][AITER] Keep all AITER kernels in `_aiter_ops` class like `_custom_ops` and `_ipex_ops` (#24490)
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>

f080a835

07 Nov, 2025 1 commit
- [Bugfix] Use latency MOE backend as default for Flashinfer and other misc fixes (#27439) · 72b1c2ae
  Pavani Majety authored Nov 07, 2025
```
Signed-off-by: Pavani Majety <pmajety@nvidia.com>
```
  72b1c2ae
04 Nov, 2025 1 commit
- [Kernels] Isolate modular kernel code from FusedMoEMethodBase subclasses. (#27123) · 938772af
  bnellnm authored Nov 04, 2025
  
  938772af
30 Oct, 2025 1 commit

[Feature] Batch invariant torch.compile (#27660) · e7acb200

Paul Zhang authored Oct 30, 2025


Signed-off-by: PaulZhang12 <paulzhan@fb.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>

e7acb200

22 Oct, 2025 1 commit

[Chore] Separate out optional dependency checks from vllm.utils (#27207) · 3ae082c3

dongbo910220 authored Oct 22, 2025

Signed-off-by: dongbo910220 <1275604947@qq.com>
Signed-off-by: dongbo910220 <32610838+dongbo910220@users.noreply.github.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>

3ae082c3

18 Oct, 2025 1 commit
- [Feature] Batch Invariant: Support DeepGEMM and Blackwell (#27127) · 245e4f2c
  Wentao Ye authored Oct 18, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  245e4f2c
16 Oct, 2025 3 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

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

[ROCm][FEAT] Fuse DeepSeek shared experts into AITER fused_moe ops (#24097) · 13170343

kliuae authored Oct 16, 2025


Signed-off-by: chenjun <junchen2@amd.com>
Signed-off-by: kliuae <kuanfu.liu@embeddedllm.com>
Co-authored-by: valarLip <103567126+valarLip@users.noreply.github.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>

13170343

14 Oct, 2025 1 commit

[torch.compile] Unwrap fused_marlin_moe custom op (#26739) · 8ae16928

Varun Sundar Rabindranath authored Oct 13, 2025


Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>

8ae16928

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
09 Oct, 2025 1 commit
- [Model] Apply shared experts overlap optimization to all models with shared experts (#26145) · 47e66c24
  bnellnm authored Oct 09, 2025
```
Signed-off-by: Bill Nell <bnell@redhat.com>
```
  47e66c24
05 Oct, 2025 2 commits
- Convert formatting to use `ruff` instead of `yapf` + `isort` (#26247) · d6953beb
  Harry Mellor authored Oct 05, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  d6953beb
- [Bugfix]: Assertion error when using FlashInfer backend (#25933) · 432e1cbc
  Simon Danielsson authored Oct 05, 2025
```
Signed-off-by: simondanielsson <simon.danielsson99@hotmail.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
```
  432e1cbc
03 Oct, 2025 1 commit
- [Refactor] Optimize FP8 MOE Backend Choice and Log (#26044) · c1ffcb55
  Wentao Ye authored Oct 03, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  c1ffcb55
02 Oct, 2025 1 commit

[Perf] Fix and reapply move apply w8a8 block fp8 linear to class (#25696) · 502640c3

ElizaWszola authored Oct 02, 2025


Signed-off-by: ElizaWszola <ewszola@redhat.com>
Signed-off-by: ElizaWszola <elizaw.9289@gmail.com>
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Signed-off-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Luka Govedič <lgovedic@redhat.com>

502640c3

01 Oct, 2025 1 commit
- [Log] Optimize Log for FP8MOE (#25709) · 12449488
  Wentao Ye authored Sep 30, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  12449488
30 Sep, 2025 1 commit
- [Bugfix] Fix accuracy issue of TRTLLM FP8 MOE and improve logging (#25895) · ef283548
  Pavani Majety authored Sep 30, 2025
```
Signed-off-by: Pavani Majety <pmajety@nvidia.com>
```
  ef283548
25 Sep, 2025 2 commits
- Revert "[Performance] Move apply_w8a8_block_fp8_linear to an op class… (#25607) · 1260180c
  Tyler Michael Smith authored Sep 25, 2025
```
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
```
  1260180c
- [Model] Add LongCat-Flash (#23991) · 845adb3e
  XuruiYang authored Sep 25, 2025
```
Signed-off-by: yangxurui <yangxurui@meituan.com>
Co-authored-by: yangxurui <yangxurui@meituan.com>
```
  845adb3e
24 Sep, 2025 1 commit
- [Refactor] Use DeepGEMM Col Major TMA Aligned Tensor (#25517) · 1f291412
  Wentao Ye authored Sep 24, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  1f291412
23 Sep, 2025 2 commits

[Performance] Move apply_w8a8_block_fp8_linear to an op class (#24666) · 63400259

ElizaWszola authored Sep 23, 2025


Signed-off-by: ElizaWszola <ewszola@redhat.com>
Signed-off-by: ElizaWszola <elizaw.9289@gmail.com>
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Signed-off-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Luka Govedič <lgovedic@redhat.com>

63400259

[Kernels] Support blocked fp8 quantization for compressed tensors MoE (#25219) · f11e3c51
bnellnm authored Sep 23, 2025
```
Signed-off-by: Bill Nell <bnell@redhat.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
```
f11e3c51

18 Sep, 2025 1 commit
- Refactor dense FP8 tensor/channel/block utils and add CT FP8 block (#21404) · fbd6523a
  Michael Goin authored Sep 18, 2025
  
  fbd6523a
17 Sep, 2025 1 commit
- [Kernel] Delegate construction of FusedMoEQuantConfig to FusedMoEMethodBase subclasses (#22537) · 5963b98b
  bnellnm authored Sep 17, 2025
```
Signed-off-by: Bill Nell <bnell@redhat.com>
```
  5963b98b
16 Sep, 2025 1 commit

[Perf] Reuse workspace for FP8+FP4 Marlin MoE (#20500) · dbebb7f8

Michael Goin authored Sep 16, 2025


Signed-off-by: mgoin <mgoin64@gmail.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>

dbebb7f8

14 Sep, 2025 1 commit

[Perf] Fix DeepGEMM Contiguous Layout Issue, 5.5% Throughput Improvement (#24783) · fc2dbcda

Wentao Ye authored Sep 14, 2025


Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>

fc2dbcda

11 Sep, 2025 3 commits
- [Bug] Fix Layer `weight_block_size` Assertion Issue (#24674) · fcba05c4
  Wentao Ye authored Sep 11, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  fcba05c4
- [Perf] Use upstream CUTLASS for SM90 Block FP8 kernel (#23280) · c3aea10d
  Michael Goin authored Sep 11, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
```
  c3aea10d
- [Doc] Remove Useless Comments (#24687) · a892b259
  Wentao Ye authored Sep 11, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  a892b259