Commits · 254a7f8fd613d6b6964abc277b73ca1f0b823cdb · OpenDAS / vllm_cscc

16 Dec, 2025 5 commits
- [Perf] Do FP4 quant before All gather on flashinfer trtllmgen MOE (#30014) · 254a7f8f
  jiahanc authored Dec 16, 2025
```
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
```
  254a7f8f
- [CI] Generalize gsm8k test args and add Qwen3-Next MTP B200 test (#30723) · 10ee1c64
  Michael Goin authored Dec 16, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  10ee1c64
- [TRTLLM] Remove the MoE GEMM weight name change (#30713) · ce12b407
  Ming Yang authored Dec 16, 2025
```
Signed-off-by: Ming Yang <minos.future@gmail.com>
```
  ce12b407
- [Feat] Enable eplb with default all2all backend (#30559) · 59bd5f6a
  Wentao Ye authored Dec 16, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  59bd5f6a
- [CustomOp] Extract ApplyRotaryEmb as CustomOp and unify the dispatch logic (#29873) · 3bd9c491
  Shanshan Shen authored Dec 16, 2025
```
Signed-off-by: shen-shanshan <467638484@qq.com>
Co-authored-by: gcanlin <canlinguosdu@gmail.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
```
  3bd9c491
15 Dec, 2025 3 commits

Update batch invariant to use attention config (#30704) · 60dbf7d8

Matthew Bonanni authored Dec 15, 2025


Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

60dbf7d8

[MoE][Refactor 1/N] Separate Online Quantization (#30627) · d0502b49
Robert Shaw authored Dec 15, 2025
```
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
```
d0502b49

[Feat] Refactor for `parallel_config` in `FusedMoEModularKernel` (#30282) · 3778673e

Wentao Ye authored Dec 14, 2025


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

3778673e

14 Dec, 2025 3 commits

[Bugfix] fix _get_quant_method of FusedMoE for deepseekV3.2 on non-NV… (#30057) · 6ecc1e41
tjp_zju authored Dec 14, 2025
```
Signed-off-by: tjp_zju <tanjianpingzju1990@gmail.com>
```
6ecc1e41

Nvidia ModelOpt workaround for issue 28072 (#30164) · 0bb0bae4

Shengliang Xu authored Dec 14, 2025


Signed-off-by: Shengliang Xu <shengliangx@nvidia.com>
Co-authored-by: Pavani Majety <pmajety@nvidia.com>

0bb0bae4

[Doc]: fixing typos in various files (#30540) · 1a55cfaf

Didier Durand authored Dec 14, 2025

Signed-off-by: Didier Durand <durand.didier@gmail.com>
Signed-off-by: Didier Durand <2927957+didier-durand@users.noreply.github.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>

1a55cfaf

13 Dec, 2025 3 commits

[Logs] Optimize startup logs 4 (#29903) · 6e78ed6b

Wentao Ye authored Dec 13, 2025

Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

6e78ed6b

[Misc][Quantization] Clarify the intent of GGUF `FusedMoE` weight materialization (#30310) · fdc135d7
Tsukasa OI authored Dec 13, 2025
```
Signed-off-by: Tsukasa OI <floss_llm@irq.a4lg.com>
```
fdc135d7

[Feature] Add SM103 (Blackwell Ultra) Support to vLLM (#30484) · 4fa7ce46

Roberto L. Castro authored Dec 13, 2025

Signed-off-by: LopezCastroRoberto <robertol.c510@gmail.com>
Signed-off-by: Roberto L. Castro <38211239+LopezCastroRoberto@users.noreply.github.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>

4fa7ce46

12 Dec, 2025 8 commits

[CI/Build][Kernel][BugFix][AMD] Fix per_token_group_quant_fp8 to use correct... · 08f8a562

rasmith authored Dec 12, 2025


[CI/Build][Kernel][BugFix][AMD] Fix per_token_group_quant_fp8 to use correct fp8 min/max values and update atol/rtol in test_quantfp8_group_functionality  (#30292)
Signed-off-by: Randall Smith <ransmith@amd.com>
Co-authored-by: Randall Smith <ransmith@amd.com>

08f8a562

[MoE-FP8-modelopt] Add FlashInfer alignment padding for intermediate dimensions (#29748) · 13618626

danielafrimi authored Dec 12, 2025


Signed-off-by: Daniel Afrimi <dafrimi@pool0-00589.cm.cluster>
Signed-off-by: dafrimi <dafrimi@nvidia.com>
Co-authored-by: Daniel Afrimi <dafrimi@pool0-00589.cm.cluster>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

13618626

[Perf] Set split_k to 1 for triton_kernels (#30528) · 1f19d8f8
Xin Yang authored Dec 12, 2025
```
Signed-off-by: Xin Yang <xyangx@amazon.com>
```
1f19d8f8
[ROCm] Enable Triton ScaledMM fallback + kernel selection fix (#26668) · cd7740ac
shivampr authored Dec 12, 2025
```
Signed-off-by: Shivam <shivampr.dev@gmail.com>
Signed-off-by: Shivam <shivamprasad91@gmail.com>
```
cd7740ac

fix(gguf): Disable bfloat16 for GGUF on blackwell device (#30408) · dc13c99e

Christina Norman authored Dec 12, 2025


Signed-off-by: Christina <truffle@gmail.com>
Signed-off-by: Isotr0py <2037008807@qq.com>
Signed-off-by: Christina Norman <christina@example.com>
Co-authored-by: Isotr0py <isotr0py@users.noreply.github.com>
Co-authored-by: Isotr0py <2037008807@qq.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

dc13c99e

[Attention] Use sparse prefill kernel for fp8 kv-cache in DeepSeek-v3.2 (#27532) · 3e41992f
Lucas Wilkinson authored Dec 12, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
3e41992f
gptq marlin quantization support for fused moe with lora (#30254) · 6a6fc41c
Bhanu Prakash Voutharoja authored Dec 12, 2025
```
Signed-off-by: Bhanu068 <voutharoja.bhanu06@gmail.com>
```
6a6fc41c
[fix] fix SM check for Flashinfer TRTLLM MOE (#30314) · 0ab23c2b
jiahanc authored Dec 11, 2025
```
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
```
0ab23c2b

11 Dec, 2025 5 commits
- [EPLB] Support EPLB w/ NVFP4 (#29804) · a00d8897
  Andrew Briand authored Dec 11, 2025
```
Signed-off-by: Andrew Briand <abriand@nvidia.com>
Co-authored-by: Andrew Briand <abriand@nvidia.com>
```
  a00d8897
- [Perf] Optimize deepgemm experts initialization, 3.9% TTFT improvement (#30494) · c817b141
  Wentao Ye authored Dec 11, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: li-jinpeng <3332126450@qq.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
```
  c817b141
- Standardise `get_rope` to use `rope_parameters["partial_rotary_factor"]`, not `rotary_dim` (#30389) · cf3eacfe
  Harry Mellor authored Dec 11, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  cf3eacfe
- [Feature] AWQ marlin quantization support for fused moe with lora (#30442) · 0e71eaa6
  汪志鹏 authored Dec 12, 2025
```
Signed-off-by: princepride <wangzhipeng628@gmail.com>
```
  0e71eaa6
- [Bugfix] Fix grouped_topk pytorch impl when num_experts can't be grouped properly (#29439) · d1e1fb43
  Divakar Verma authored Dec 10, 2025
```
Signed-off-by: Divakar Verma <divakar.verma@amd.com>
Co-authored-by: Gregory Shtrasberg <156009573+gshtras@users.noreply.github.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
```
  d1e1fb43
10 Dec, 2025 2 commits
- Fix typos in comments across multiple files (#30345) · 3bdd4266
  Wilson Wu authored Dec 10, 2025
```
Signed-off-by: Wilson Wu <iwilsonwu@gmail.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
```
  3bdd4266
- [Bugfix] Fix fp8 DeepGemm compilation issues (#30336) · 2e7035dd
  ElizaWszola authored Dec 10, 2025
  
  2e7035dd
09 Dec, 2025 11 commits

[Rocm][torch.compile] Adding layernorm + fp8 block quant and silu + fp8 block... · 3c680f4a

Charlie Fu authored Dec 09, 2025


[Rocm][torch.compile] Adding layernorm + fp8 block quant and silu + fp8 block quant for Aiter (#25693)
Signed-off-by: charlifu <charlifu@amd.com>
Signed-off-by: Micah Williamson <micah.williamson@amd.com>
Signed-off-by: Charlie Fu <Charlie.Fu@amd.com>
Co-authored-by: Micah Williamson <micah.williamson@amd.com>
Co-authored-by: wuhuikx <hattie.wu@amd.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Gregory Shtrasberg <156009573+gshtras@users.noreply.github.com>

3c680f4a

[Quantization] FP8 Weight Reloading for Quantized RL Rollout (#28480) · fccd5325
Kyle Sayers authored Dec 09, 2025
```
Signed-off-by: Kyle Sayers <kylesayrs@gmail.com>
```
fccd5325

[MoE][Refactor] Remove most arguments to FusedMoEMethodBase.apply (#29066) · 00e5cbb9

bnellnm authored Dec 09, 2025

Signed-off-by: Bill Nell <bnell@redhat.com>
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Co-authored-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>

00e5cbb9

[Feature] Batch-Invariant Support for FA2 and LoRA (#30018) · 5dcd593b

liuquan authored Dec 09, 2025


Signed-off-by: quanliu <18646313696@163.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>

5dcd593b

[ROCm] Aiter Quant Kernels (#25552) · ee14644b
vllmellm authored Dec 09, 2025
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
```
ee14644b
[Bugfix]: Fix glm46 awq marlin moe wna16 compatibility (#30210) · 1166c31c
Dongjie Zou authored Dec 09, 2025
```
Signed-off-by: baonudesifeizhai <baonudesifeizhai@gmail.com>
```
1166c31c

[Model][Quantization] Restore MoE + GGUF models support (incl. Qwen3 MoE) by... · 58d5b3f5

Tsukasa OI authored Dec 09, 2025


[Model][Quantization] Restore MoE + GGUF models support (incl. Qwen3 MoE) by allowing Sideload Parameters (#30116)
Signed-off-by: Tsukasa OI <floss_llm@irq.a4lg.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>

58d5b3f5

[Bugfix] Fix compressed-tensors models failing to load with transformers backend (#30287) · 03b91f72

Michael Goin authored Dec 08, 2025

Signed-off-by: mgoin <mgoin64@gmail.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

03b91f72

[Kernel]Support W4A8 Grouped GEMM on Hopper (#29691) · f6227c22
czhu-cohere authored Dec 08, 2025
```
Signed-off-by: czhu-cohere <conway.zhu@cohere.com>
```
f6227c22

[Bugfix] Fix DeepGEMM after #29546 (#30267) · ae339b1a

Zhewen Li authored Dec 08, 2025


Signed-off-by: zhewenli <zhewenli@meta.com>
Signed-off-by: Zhewen Li <zhewenli@meta.com>

ae339b1a

[Feature] Batch invariant: Enable `TRITON_MLA` without prefix-caching (#29125) · d9417096
Wentao Ye authored Dec 08, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
d9417096