Commits · c3270a9246c2968642ae5f3fa77002f5f42bb6f8 · OpenDAS / vllm_cscc

18 Apr, 2026 1 commit
- [feature][Attention Backend] TurboQuant: 2-bit KV cache compression with 4x capacity #38479 · 0885aa25
  wanglong3 authored Apr 18, 2026
  
  0885aa25
03 Apr, 2026 1 commit
- [BUGFIX] 回退 ROCm FlashAttention unified KV layout 改动并修正 unified kernel 选择逻辑 · b233584a
  laibao authored Apr 03, 2026
  
  b233584a
26 Mar, 2026 2 commits

refactor(rocm): 提取 unified flash 的 block_size 判定逻辑 · ee989f6d
laibao authored Mar 26, 2026

ee989f6d

feat(v1 attention): 为 ROCm FlashAttention 接入 unified kv layout，并打通... · ea9b8584

laibao authored Mar 23, 2026

feat(v1 attention): 为 ROCm FlashAttention 接入 unified kv layout，并打通 mm_prefix、qq_bias 与 use_alibi_sqrt 透传
在 ROCm FlashAttention 后端增加 unified KV layout 选择逻辑
接入 unified varlen kernel 调用路径
在 FlashAttention metadata 中补充 mm_prefix_range 与 qq_bias 透传

ea9b8584

09 Mar, 2026 1 commit
- 适配在bmz上的mla的kvcache_e5m2和e4m3量化的支持 · 0daa00fb
  yangql authored Mar 09, 2026
  
  0daa00fb
03 Mar, 2026 1 commit
- support dsa · d71496bf
  zhuwenwen authored Mar 03, 2026
  
  d71496bf
05 Feb, 2026 1 commit
- sync v0.15.1(ex fused_moe&models) · 899a2db4
  zhuwenwen authored Feb 05, 2026
  
  899a2db4
28 Jan, 2026 1 commit
- 修复了awq的shape的bug，以及兼容了lmslim注册导入的的情况 · 2d8b3257
  zhuwenwen authored Jan 28, 2026
  
  2d8b3257
27 Jan, 2026 1 commit
- [Perf] avoid duplicate mem_get_info() call in get_current_memory_usage (#33064) · 157caf51
  Paco Xu authored Jan 27, 2026
```
Signed-off-by: Paco Xu <paco.xu@daocloud.io>
```
  157caf51
24 Jan, 2026 1 commit
- [ROCm][ViT] Enable Flash Attention Triton backend on RDNA3/RDNA4 (#32944) · 97ef11dd
  monajafi-amd authored Jan 23, 2026
```
Signed-off-by: mohammad najafi <mohammad.najafi@amd.com>
```
  97ef11dd
23 Jan, 2026 1 commit
- [Refactor] Clean up unused variables & func (#32692) · 37c9859f
  Wentao Ye authored Jan 23, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  37c9859f
22 Jan, 2026 3 commits
- [Hardware][AMD][CI][Bugfix] Fix regressions from deprecated env vars (#32837) · c517d8c9
  Matt authored Jan 22, 2026
```
Signed-off-by: Matthew Wong <Matthew.Wong2@amd.com>
```
  c517d8c9
- [AMD][ROCm] MoRI EP: a high-performance all2all backend (#28664) · 49a12622
  Alex Sun authored Jan 22, 2026
```
Signed-off-by: Alex Sun <alex.s@amd.com>
```
  49a12622
- [Deprecation] Remove deprecated environment variables (#32812) · 6437ff1f
  Wentao Ye authored Jan 21, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  6437ff1f
21 Jan, 2026 1 commit
- [ROCm][Deepseekv3.2] Refactor Sparse Indexer as CustomOp (#29287) · 6c20e89c
  Pleaplusone authored Jan 21, 2026
```
Signed-off-by: ganyi <ygan@amd.com>
```
  6c20e89c
16 Jan, 2026 3 commits

set VLLM_USE_MARLIN_W16A16_MOE=0 on bw · 8cbcac5d
zhuwenwen authored Jan 16, 2026

8cbcac5d

解决custom cudagraph模式需要拷贝的问题，需要配合dtk进行使用。 · f1bc9890

zhuwenwen authored Jan 16, 2026

区分pcie和hglink custom allreduce的使用
vllm：export VLLM_CUSTOM_CACHE=1
dtk：export HIP_KERNEL_EVENT_SYSTENFENCE=1

set VLLM_USE_FUSED_RMS_ROPE=1
add SUPPORT_MOE_MARLIN_W16A16 to use moe marlin on bw
support fa kvcache fp8 (todo: add VLLM_USE_QUERY_QUANT to not use q quant)
update moe_align_block_size

f1bc9890

Switch default w8a8 gemm impl to blaslt. · f06d1125
zhuwenwen authored Jan 16, 2026
```
fix _forward_encoder_attention
remove medusa
set VLLM_PCIE_USE_CUSTOM_ALLREDUCE=1
```
f06d1125

14 Jan, 2026 1 commit
- AMD CI Test - unskip moe_sum test and moe_align_block_size tests (#32039) · 048bb597
  Hongxia Yang authored Jan 14, 2026
```
Signed-off-by: Hongxia Yang <hongxia.yang@amd.com>
```
  048bb597
11 Jan, 2026 1 commit
- [Hardware][AMD][CI][Bugfix] Fix AMD Quantization test group (#31713) · bde57ab2
  Matt authored Jan 11, 2026
```
Signed-off-by: Matthew Wong <Matthew.Wong2@amd.com>
```
  bde57ab2
09 Jan, 2026 1 commit
- [1/N][Attention] Restructure attention: move files (#31916) · 2612ba92
  Matthew Bonanni authored Jan 09, 2026
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  2612ba92
05 Jan, 2026 1 commit
- [v1] Add encoder-only/cross attention support to Triton Attention backend (#31406) · 6aa5b18e
  Isotr0py authored Jan 06, 2026
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  6aa5b18e
30 Dec, 2025 1 commit

[ROCm][Bugfix] Fix accuracy issue on fmoe when... · 1a834df2

Pleaplusone authored Dec 30, 2025


[ROCm][Bugfix] Fix accuracy issue on fmoe when `VLLM_ROCM_USE_AITER_FUSION_SHARED_EXPERTS` enabled (#31523)
Signed-off-by: ganyi <ygan@amd.com>

1a834df2

25 Dec, 2025 1 commit
- fix run error · 1217257c
  zhuwenwen authored Dec 25, 2025
  
  1217257c
22 Dec, 2025 1 commit
- [Bugfix][ROCm] Fix typo: is_linear_fp8_enaled -> is_linear_fp8_enabled (#31109) · cf8eed7b
  Kevin McKay authored Dec 21, 2025
```
Signed-off-by: c0de128 <kevin.mckay@outlook.com>
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
```
  cf8eed7b
19 Dec, 2025 2 commits
- remove unused code · de5774fa
  zhuwenwen authored Dec 19, 2025
  
  de5774fa
- [ROCm][CI][Bugfix] Multi-Modal Model Support Fixes and Attention Backend Improvements (#30270) · 7b43db21
  Andreas Karatzas authored Dec 18, 2025
```
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
```
  7b43db21
16 Dec, 2025 1 commit
- [ROCm] [AITER] [DOC] Add usage description about check functions in `_aiter_ops` (#30586) · d0fb5729
  TJian authored Dec 16, 2025
```
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
```
  d0fb5729
15 Dec, 2025 2 commits

[Platform] Refactor Platform attention backend selection to avoid breakpoint... · ec154c36

Isotr0py authored Dec 16, 2025


[Platform] Refactor Platform attention backend selection to avoid breakpoint for OOT platform (#30212)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Signed-off-by: Isotr0py <2037008807@qq.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

ec154c36

[CustomOp][MM] Extract MMEncoderAttention as CustomOp and replace the backend... · 87b4d155

Shanshan Shen authored Dec 15, 2025


[CustomOp][MM] Extract MMEncoderAttention as CustomOp and replace the backend of QwenVisionAttention with it. (#30125)
Signed-off-by: shen-shanshan <467638484@qq.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: tjtanaa <tunjian.tan@embeddedllm.com>

87b4d155

13 Dec, 2025 1 commit
- update deps，fix import and optional error, remove tc · cffe15ef
  zhuwenwen authored Dec 13, 2025
  
  cffe15ef
11 Dec, 2025 1 commit
- [ROCm] Fix broken import in platform attention backend dispatching (#30432) · b51255f3
  Andreas Karatzas authored Dec 10, 2025
```
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
```
  b51255f3
09 Dec, 2025 1 commit
- [ROCm] Aiter Quant Kernels (#25552) · ee14644b
  vllmellm authored Dec 09, 2025
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
```
  ee14644b
07 Dec, 2025 1 commit
- [v1] Add PrefixLM support to FlexAttention backend (#27938) · b952f4d3
  Isotr0py authored Dec 07, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  b952f4d3
05 Dec, 2025 1 commit
- [PCP&DCP] move CUDAGraph check for PCP&DCP to the check func of platforms (#29952) · 0098a6e3
  Qiu authored Dec 05, 2025
```
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
```
  0098a6e3
27 Nov, 2025 1 commit
- [Attention] Update attention imports (#29540) · fc1d8be3
  Matthew Bonanni authored Nov 27, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  fc1d8be3
25 Nov, 2025 2 commits
- [ROCm][CI] Fix test_cudagraph_mode failure in AMD CI (#29367) · ef1f7030
  Micah Williamson authored Nov 25, 2025
```
Signed-off-by: Micah Williamson <micah.williamson@amd.com>
```
  ef1f7030
- [Bugfix] [ROCm] [UX]: revert Flex attention backend (#29371) · 64deead7
  vllmellm authored Nov 25, 2025
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
```
  64deead7
24 Nov, 2025 1 commit
- [Bugfix] [ROCm] [UX] Reorganize ROCm Backend Selection Logic (#26980) · e48b2e68
  vllmellm authored Nov 24, 2025
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
```
  e48b2e68
21 Nov, 2025 1 commit
- [Attention] Add ROCM_AITER_MLA_SPARSE to attention backend registry (#29103) · 11857a00
  Matthew Bonanni authored Nov 20, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  11857a00