Commits · 69f30ae059e7623918637336457f1be42c8652c0 · OpenDAS / vllm_cscc

01 Sep, 2025 1 commit
- add dca and sparse attention support on rocm · 12112a4d
  zhuwenwen authored Sep 01, 2025
  
  12112a4d
28 Aug, 2025 1 commit
- 优化 block_tables 的生成逻辑，增加对混合情况的检测，确保在存在空和非空块时正确计算最大块长度。 · 4932783d
  laibao authored Aug 28, 2025
  
  4932783d
21 Aug, 2025 1 commit
- revert 修改flashmla的接口 · 9b67a1e0
  zhuwenwen authored Aug 21, 2025
  
  9b67a1e0
19 Aug, 2025 1 commit
- update v1 fa layout · 864c718a
  zhuwenwen authored Aug 19, 2025
  
  864c718a
15 Aug, 2025 2 commits
- 修改flashmla的接口 · 299df36d
  yangql authored Aug 15, 2025
  
  299df36d
- 增加marlin对cache13的支持，以及新增flash mla的kvcache fp8的支持 · 8f25283a
  yangql authored Aug 15, 2025
  
  8f25283a
06 Aug, 2025 1 commit
- update VLLM_FLASH_ATTN_V1 to VLLM_USE_FLASH_ATTN_PA · 88dbf92c
  zhuwenwen authored Aug 06, 2025
  
  88dbf92c
05 Aug, 2025 1 commit
- when using VLLM_FLASH_ATTN_V1, set block_size to 64 · 80a682c7
  zhuwenwen authored Aug 05, 2025
  
  80a682c7
24 Jul, 2025 1 commit
- update fa v1 interface · f9408aff
  zhuwenwen authored Jul 24, 2025
  
  f9408aff
17 Jul, 2025 3 commits
- update adpaptive block_size · 937a3ec1
  zhuwenwen authored Jul 17, 2025
  
  937a3ec1
- set default block_size and pa · b61f7a69
  zhuwenwen authored Jul 17, 2025
  
  b61f7a69
- add VLLM_USE_PA_PRINT_PARAM to print fa-pa size · 1092a467
  zhuwenwen authored Jul 17, 2025
  
  1092a467
10 Jul, 2025 1 commit
- update bw vdim128 · d805c59c
  zhuwenwen authored Jul 10, 2025
  
  d805c59c
09 Jul, 2025 1 commit
- [fix]修复零消耗引入的cudagraph模式数据准备cpu耗时问题 · cd6239dc
  zhuwenwen authored Jul 09, 2025
  
  cd6239dc
06 Jul, 2025 3 commits
- Revert "[V0 deprecation] Remove V0 CPU/XPU/TPU backends (#20412)" · a5dd03c1
  simon-mo authored Jul 06, 2025
```
This reverts commit e202dd27.
```
  a5dd03c1
- [V1] Support any head size for FlexAttention backend (#20467) · 9fb52e52
  Cyrus Leung authored Jul 07, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  9fb52e52
- [V0 deprecation] Remove V0 CPU/XPU/TPU backends (#20412) · e202dd27
  Woosuk Kwon authored Jul 06, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
Signed-off-by: jiang1.li <jiang1.li@intel.com>
Co-authored-by: Li, Jiang <jiang1.li@intel.com>
```
  e202dd27
03 Jul, 2025 1 commit
- update VLLM_USE_FLASH_ATTN_BACKEND to VLLM_USE_FLASH_ATTN_PA · e036115e
  zhuwenwen authored Jul 03, 2025
  
  e036115e
02 Jul, 2025 1 commit
- [TPU] kv cache update kernel supports dynamic grid (#20235) · 7da296be
  Chengji Yao authored Jul 01, 2025
```
Signed-off-by: Chengji Yao <chengjiyao@google.com>
```
  7da296be
27 Jun, 2025 2 commits
- update cutlass fa and pa · 4ddb5447
  zhuwenwen authored Jun 27, 2025
  
  4ddb5447
- Quick Fix by adding conditional import for flash_attn_varlen_func in flash_attn (#20143) · dec197e3
  Chendi.Xue authored Jun 27, 2025
```
Signed-off-by: Chendi.Xue <chendi.xue@intel.com>
```
  dec197e3
26 Jun, 2025 3 commits
- [TPU] add kv cache update kernel (#19928) · 04e1642e
  Chengji Yao authored Jun 26, 2025
```
Signed-off-by: Chengji Yao <chengjiyao@google.com>
```
  04e1642e
- [Hardware][Intel GPU] Add v1 Intel GPU support with Flash attention backend. (#19560) · b69781f1
  Kunshang Ji authored Jun 27, 2025
```
Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
```
  b69781f1
- [Bugfix][V1][ROCm] Fix AITER Flash Attention Backend (Fix API Break and Local... · 27c065df
  TJian authored Jun 26, 2025
```
[Bugfix][V1][ROCm] Fix AITER Flash Attention Backend (Fix API Break and Local Attention Logic: affecting Llama4) (#19904)
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
```
  27c065df
25 Jun, 2025 1 commit
- [Refactor] Remove duplicate `ceil_div` (#20023) · 879f69be
  Wentao Ye authored Jun 25, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  879f69be
23 Jun, 2025 1 commit
- update common.py · 2da6fd72
  zhuwenwen authored Jun 23, 2025
  
  2da6fd72
21 Jun, 2025 1 commit
- only K100_AI can use cutlass prefix cache · 14945681
  zhuwenwen authored Jun 21, 2025
  
  14945681
20 Jun, 2025 2 commits
- [Kernel] mark TorchSDPABackend swap_blocks NotImplementedError (#19749) · 71baf85a
  Ning Xie authored Jun 21, 2025
  
  71baf85a
- [Kernel] correct cpu worker function parameter type (#19745) · 71d12195
  Ning Xie authored Jun 20, 2025
```
Signed-off-by: Andy Xie <andy.xning@gmail.com>
```
  71d12195
18 Jun, 2025 3 commits
- [Minor] Zero-initialize attn output buffer (#19784) · f04d6045
  Woosuk Kwon authored Jun 17, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  f04d6045
- release opt1 · 1a280da1
  zhuwenwen authored Jun 18, 2025
  
  1a280da1
- [MISC] correct copy_blocks src_to_dists param type (#19696) · c53711bd
  Ning Xie authored Jun 18, 2025
```
Signed-off-by: Andy Xie <andy.xning@gmail.com>
```
  c53711bd
17 Jun, 2025 2 commits
- [V1][Kernel] Flashinfer HND KV cache layout (#19280) · 4c8f64fa
  Nicolò Lucchesi authored Jun 17, 2025
```
Signed-off-by: NickLucche <nlucches@redhat.com>
```
  4c8f64fa
- [Kernel] Add Split-KV Support to Unified Triton Attention Kernel (#19152) · ccd7c050
  jvlunteren authored Jun 17, 2025
```
Signed-off-by: Jan van Lunteren <jvl@zurich.ibm.com>
```
  ccd7c050
15 Jun, 2025 1 commit
- [Kernel] Raise verbose error and consolidate `num_heads/num_kv_heads` divisibility check (#19339) · 0b73736a
  22quinn authored Jun 14, 2025
```
Signed-off-by: 22quinn <33176974+22quinn@users.noreply.github.com>
```
  0b73736a
13 Jun, 2025 1 commit
- update triton_mla.py · e046b382
  zhuwenwen authored Jun 13, 2025
  
  e046b382
12 Jun, 2025 2 commits
- [torch.compile][ROCm] Fuse quantization onto attention using a torch.compile pass (#16756) · f98548b9
  Luka Govedič authored Jun 12, 2025
```
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Co-authored-by: Sage Moore <sage@neuralmagic.com>
```
  f98548b9
- [CI] change spell checker from codespell to typos (#18711) · 2f1c19b2
  Ning Xie authored Jun 12, 2025
```
Signed-off-by: Andy Xie <andy.xning@gmail.com>
```
  2f1c19b2
11 Jun, 2025 2 commits

[AMD] [Quantization] Add override flag for attention dtype instead of using... · c7ea0b56

rasmith authored Jun 11, 2025


[AMD] [Quantization] Add override flag for attention dtype instead of using kv_cache_dtype trigger (#17331)
Signed-off-by: Randall Smith <Randall.Smith@amd.com>

c7ea0b56

[Misc] Fix misleading ROCm warning (#19486) · 04a55612
Jee Jee Li authored Jun 12, 2025
```
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
```
04a55612