Commits · 7a985548488ddbfa965f9684acf6f1d53f905964 · OpenDAS / vllm_cscc

15 May, 2025 2 commits
- [Kernel] [V1] Fix performance regression for triton unified attention (#18161) · 01c22335
  Thomas Parnell authored May 15, 2025
```
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Co-authored-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  01c22335
- [Bugfix] Fix fp8 tests for triton_unified_attention for Triton 3.3 (#18013) · e6b8e65d
  Thomas Parnell authored May 15, 2025
```
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Co-authored-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  e6b8e65d
14 May, 2025 1 commit
- [BugFix][AMD] Compatible patch for AITER lib after 04/20 (#17912) · 4f8b3732
  qli88 authored May 14, 2025
```
Signed-off-by: Qiang Li <qiang.li2@amd.com>
```
  4f8b3732
11 May, 2025 1 commit
- [FP8][ROCm][Attention] Enable FP8 KV cache on ROCm for V1 (#17870) · 06c0922a
  Gregory Shtrasberg authored May 11, 2025
```
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>
```
  06c0922a
09 May, 2025 3 commits
- Revert "[BugFix][AMD] Compatible patch for latest AITER(05/07/2025)" (#17910) · 85b72cb7
  Michael Goin authored May 09, 2025
  
  85b72cb7
- [BugFix][AMD] Compatible patch for latest AITER(05/07/2025) (#17864) · 9f64e934
  qli88 authored May 09, 2025
```
Signed-off-by: Qiang Li <qiang.li2@amd.com>
```
  9f64e934
- [FEAT][ROCm]: Support AITER MLA on V1 Engine (#17523) · 3c9396a6
  vllmellm authored May 09, 2025
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
Co-authored-by: qli88 <qiang.li2@amd.com>
Co-authored-by: Hongxia Yang <62075498+hongxiayang@users.noreply.github.com>
```
  3c9396a6
08 May, 2025 1 commit
- [Hardware][Intel-Gaudi] Support Automatic Prefix Caching on HPU (#17648) · 843b2227
  Agata Dobrzyniewicz authored May 08, 2025
```
Signed-off-by: Agata Dobrzyniewicz <adobrzyniewicz@habana.ai>
```
  843b2227
06 May, 2025 2 commits
- [Kernel] Unified Triton kernel that doesn't distinguish between prefill + decode (#16828) · 2f925e57
  Thomas Parnell authored May 06, 2025
```
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Co-authored-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  2f925e57
- [Bugfix] Fix triton import with local TritonPlaceholder (#17446) · f9bc5a06
  Mengqing Cao authored May 06, 2025
```
Signed-off-by: Mengqing Cao <cmq0113@163.com>
```
  f9bc5a06
02 May, 2025 1 commit
- [Bugfix] fix tmp_out and exp_sums dimensions (#17438) · 4c33d673
  Hui Liu authored May 02, 2025
```
Signed-off-by: Hui Liu <96135754+hliuca@users.noreply.github.com>
```
  4c33d673
01 May, 2025 1 commit
- [ROCm] remove unsupported archs from rocm triton flash-attention supported list (#17536) · 28566d73
  Hongxia Yang authored May 01, 2025
```
Signed-off-by: Hongxia Yang <hongxia.yang@amd.com>
```
  28566d73
30 Apr, 2025 1 commit
- Update PyTorch to 2.7.0 (#16859) · 2c4f59af
  Huy Do authored Apr 29, 2025
  
  2c4f59af
27 Apr, 2025 1 commit

[Kernel][Triton][FP8] Adding fp8 and variable length sequence support to... · 8e4b351a

rasmith authored Apr 26, 2025


[Kernel][Triton][FP8] Adding fp8 and variable length sequence support to Triton FAv2 kernel (#12591)
Signed-off-by: Randall Smith <Randall.Smith@amd.com>

8e4b351a

26 Apr, 2025 1 commit
- [Hardware][Intel-Gaudi] Update hpu-extension and update bucketing system for HPU device (#17186) · c48334d4
  Agata Dobrzyniewicz authored Apr 26, 2025
```
Signed-off-by: Agata Dobrzyniewicz <adobrzyniewicz@habana.ai>
```
  c48334d4
23 Apr, 2025 1 commit

[Kernel][ROCM] Upstream prefix prefill speed up for vLLM V1 (#13305) · bc7c4d20

Aleksandr Malyshev authored Apr 22, 2025


Signed-off-by: Sage Moore <sage@neuralmagic.com>
Signed-off-by: root <root@banff-cyxtera-s73-5.ctr.dcgpu>
Signed-off-by: Aleksandr Malyshev <maleksan@amd.com>
Signed-off-by: root <root@banff-cyxtera-s65-4.amd.com>
Signed-off-by: maleksan85 <maleksan@amd.com>
Signed-off-by: <>
Co-authored-by: Sage Moore <sage@neuralmagic.com>
Co-authored-by: root <root@banff-cyxtera-s73-5.ctr.dcgpu>
Co-authored-by: Aleksandr Malyshev <maleksan@amd.com>
Co-authored-by: qli88 <qiang.li2@amd.com>
Co-authored-by: root <root@banff-cyxtera-s65-4.amd.com>

bc7c4d20

22 Apr, 2025 2 commits

[FEAT][ROCm]: Support AITER MLA (#15893) · 30bc3e0f

vllmellm authored Apr 23, 2025


Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
Co-authored-by: qli88 <qiang.li2@amd.com>

30bc3e0f

[FEAT][ROCm] Integrate Paged Attention Kernel from AITER (#15001) · 0e237f00

vllmellm authored Apr 22, 2025


Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
Co-authored-by: tjtanaa <tunjian.tan@embeddedllm.com>

0e237f00

15 Apr, 2025 1 commit
- [Kernel] Remove redundant Exp calculations (#16123) · 280d62b8
  DefTruth authored Apr 15, 2025
```
Signed-off-by: DefTruth <qiustudent_r@163.com>
```
  280d62b8
14 Apr, 2025 1 commit
- [Misc] remove warning if triton>=3.2.0 (#16553) · ce4ddd2d
  DefTruth authored Apr 14, 2025
```
Signed-off-by: DefTruth <qiustudent_r@163.com>
```
  ce4ddd2d
11 Apr, 2025 1 commit
- [Kernel] support merge_attn_states CUDA kernel, 3x speedup (#16173) · e9528f6d
  DefTruth authored Apr 11, 2025
```
Signed-off-by: DefTruth <qiustudent_r@163.com>
```
  e9528f6d
09 Apr, 2025 1 commit
- [BugFix] fix some typos found by typos. (#16314) · 04149cce
  yihong authored Apr 09, 2025
```
Signed-off-by: yihong0618 <zouzou0208@gmail.com>
```
  04149cce
04 Apr, 2025 1 commit
- [ROCm][Bugfix] Use platform specific FP8 dtype (#15717) · 40a36ccf
  Gregory Shtrasberg authored Apr 04, 2025
```
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>
```
  40a36ccf
03 Apr, 2025 2 commits

[Neuron][kernel] Fuse kv cache into a single tensor (#15911) · d2b58ca2
Liangfu Chen authored Apr 03, 2025
```
Signed-off-by: Liangfu Chen <liangfc@amazon.com>
```
d2b58ca2

[ROCM][KERNEL] Paged attention for V1 (#15720) · e73ff24e

Aleksandr Malyshev authored Apr 02, 2025


Signed-off-by: Aleksandr Malyshev <maleksan@amd.com>
Signed-off-by: root <root@banff-cyxtera-s65-4.amd.com>
Co-authored-by: Aleksandr Malyshev <maleksan@amd.com>
Co-authored-by: root <root@banff-cyxtera-s65-4.amd.com>

e73ff24e

26 Mar, 2025 1 commit

[BugFix] Fix nightly MLA failure (FA2 + MLA chunked prefill, i.e. V1,... · 33437bc6

Lucas Wilkinson authored Mar 25, 2025


[BugFix] Fix nightly MLA failure (FA2 + MLA chunked prefill, i.e. V1, producing bad results) (#15492)
Signed-off-by: LucasWilkinson <lwilkinson@neuralmagic.com>

33437bc6

14 Mar, 2025 1 commit

[Kernel] [V1] Further optimizations to ROCm (Triton) Backend to better handle GQA. (#14431) · fb4c7f8e

Thomas Parnell authored Mar 14, 2025


Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Co-authored-by: Jan van Lunteren <jvl@zurich.ibm.com>
Co-authored-by: Burkhard Ringlein <ngl@zurich.ibm.com>
Co-authored-by: Chih-Chieh Yang <chih.chieh.yang@ibm.com>

fb4c7f8e

12 Mar, 2025 1 commit

[CPU] Upgrade CPU backend to torch-2.6 (#13381) · ff47aab0

Li, Jiang authored Mar 12, 2025


Signed-off-by: jiang1.li <jiang1.li@intel.com>
Co-authored-by: Isotr0py <2037008807@qq.com>

ff47aab0

11 Mar, 2025 1 commit
- [neuron] add reshape_and_cache (#14391) · c91b64f7
  Liangfu Chen authored Mar 10, 2025
  
  c91b64f7
06 Mar, 2025 2 commits
- Add authors to license header. (#14371) · e642ec96
  Thomas Parnell authored Mar 06, 2025
```
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Co-authored-by: Burkhard Ringlein <ngl@zurich.ibm.com>
Co-authored-by: Jan van Lunteren <jvl@zurich.ibm.com>
```
  e642ec96
- [Kernel] [V1] Improved performance for V1 Triton (ROCm) backend (#14152) · 6bd1dd9d
  Thomas Parnell authored Mar 06, 2025
  
  6bd1dd9d
27 Feb, 2025 2 commits
- [core] Perf improvement for DSv3 on AMD GPUs (#13718) · 8294773e
  qli88 authored Feb 27, 2025
```
Signed-off-by: qli88 <qiang.li2@amd.com>
```
  8294773e
- [Kernel] FlashMLA integration (#13747) · f9590390
  Lucas Wilkinson authored Feb 26, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  f9590390
22 Feb, 2025 1 commit
- [V1][Kernel] Refactor the prefix_prefill kernel so that the caller no longer... · 558db808
  Sage Moore authored Feb 22, 2025
```
[V1][Kernel] Refactor the prefix_prefill kernel so that the caller no longer has to pass in the context lengths (#13095)
```
  558db808
21 Feb, 2025 2 commits

[Attention] MLA with chunked prefill (#12639) · 288cc6c2

Lucas Wilkinson authored Feb 21, 2025


Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Patrick Horn <patrick.horn@gmail.com>
Co-authored-by: simon-mo <xmo@berkeley.edu>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

288cc6c2

[Neuron][Kernel] Vectorize KV cache load in FlashPagedAttention to maximize DMA bandwidth (#13245) · 33170081
Lingfan Yu authored Feb 20, 2025
```
Signed-off-by: Lingfan Yu <lingfany@amazon.com>
```
33170081

19 Feb, 2025 1 commit

[Hardware][Gaudi][Feature] Support Contiguous Cache Fetch (#12139) · d0a7a276

Yu-Zhou authored Feb 19, 2025


Signed-off-by: yuzhou <yuzhou@habana.ai>
Signed-off-by: zhouyu5 <yu.zhou@intel.com>
Co-authored-by: Cody Yu <hao.yu.cody@gmail.com>

d0a7a276

14 Feb, 2025 1 commit
- [ROCm][V1] Add intial ROCm support to V1 (#12790) · ba59b78a
  Sage Moore authored Feb 13, 2025
  
  ba59b78a
12 Feb, 2025 1 commit

[Neuron][Kernel] Support Longer Sequences in NKI-based Flash PagedAttention... · e92694b6

Lingfan Yu authored Feb 11, 2025


[Neuron][Kernel] Support Longer Sequences in NKI-based Flash PagedAttention and Improve Efficiency (#12921)
Signed-off-by: Lingfan Yu <lingfany@amazon.com>

e92694b6

05 Feb, 2025 1 commit
- [ROCM][AMD][TRITON] Halving warps number for fw_prefill to reduce spilling (#12713) · 64862d10
  Aleksandr Malyshev authored Feb 04, 2025
```
Signed-off-by: Aleksandr Malyshev <maleksan@amd.com>
Co-authored-by: Aleksandr Malyshev <maleksan@amd.com>
```
  64862d10