a97bd607af
[Core] Add Prefill Context Parallelism (PCP) and All-to-All DCP communication
...
This PR adds Prefill Context Parallelism (PCP) support for splitting prefill
tokens across ranks using a DualChunkSwap pattern, and integrates an All-to-All
communication backend for Decode Context Parallelism (DCP).
Key changes:
- Add PCP with DualChunkSwap token partitioning for balanced prefill computation
- Add All-to-All DCP communication backend reducing NCCL calls from 3 to 2
- Restrict DCP+PCP to two clean configurations:
- Case 1: DCP = PCP (same TP position, all-reduce only)
- Case 2: DCP = TP × PCP (full TP all-gather, all-reduce + slice)
- Add PCPManager for buffer management and input partitioning
- Update attention backends (FlashAttention, FlashInfer, MLA) for PCP support
- Add comprehensive tests for DCP operations
Co-Authored-By: QiuChunshuo <qiuchunshuo@huawei.com >
Co-Authored-By: zhenwenqi2024 <zhenwenqi_2022@qq.com >
Co-Authored-By: FENP <yuanyongjie.yyj@antgroup.com >
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com >
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com >
2026-03-02 05:40:48 +00:00
Matthew Bonanni and GitHub
7eb6cb6c18
[Attention] Update tests to remove deprecated env vars ( #30563 )
...
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com >
2025-12-17 09:49:59 -08:00
Jaya Yuan and GitHub
67475a6e81
[DCP][Bugfix][CI] Fix accuracy issue of DCP when using FLASH_ATTN_MLA ( #30309 )
...
Signed-off-by: FENP <yuanyongjie.yyj@antgroup.com >
2025-12-09 08:22:14 +00:00
Qiu and GitHub
46cbbca05c
[CI][DCP][Perf] reduce DCP CI execution time ( #29858 )
...
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com >
2025-12-04 17:28:21 +00:00
2fd893b4ce
[Feature] Prefill Context Parallel (PCP) basic support ( #28718 )
...
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com >
Signed-off-by: FENP <yuanyongjie.yyj@antgroup.com >
Signed-off-by: LookAround <lixushi@huawei.com >
Signed-off-by: Jingchun Gao <gaojingchun1@huawei.com >
Signed-off-by: zhenwenqi2024 <zhenwenqi_2022@qq.com >
Co-authored-by: FENP <yuanyongjie.yyj@antgroup.com >
Co-authored-by: LookAround <lixushi@huawei.com >
Co-authored-by: Jingchun Gao <gaojingchun1@huawei.com >
Co-authored-by: zhenwenqi2024 <zhenwenqi_2022@qq.com >
Co-authored-by: Jingchun Gao <63247409+gjc0824@users.noreply.github.com >
2025-11-19 15:52:44 -05:00
4516d44b7f
[DCP] Support Decode Context Parallel (DCP) for GQA with Flashinfer ( #25438 )
...
Signed-off-by: gaojc <1055866782@qq.com >
Signed-off-by: Jingchun Gao <gaojingchun1@huawei.com >
Signed-off-by: Jingchun Gao <63247409+gjc0824@users.noreply.github.com >
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com >
Co-authored-by: gaojingchun (A) <g00955623@china.huawei.com >
Co-authored-by: Jingchun Gao <gaojingchun1@huawei.com >
Co-authored-by: QiuChunshuo <qiuchunshuo@huawei.com >
2025-11-14 11:24:10 +00:00
39029d5192
[CI/Test Fix] Fix CP tests on Blackwell ( #28404 )
...
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com >
Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com >
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2025-11-11 01:36:29 +00:00
2108a571d7
[DCP] Support dcp kv_cache interleave size > 1 ( #26696 )
...
Signed-off-by: zhangsicheng5 <zhangsicheng5@huawei.com >
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com >
Signed-off-by: Qiu <qiuchunshuo@huawei.com >
Co-authored-by: QiuChunshuo <qiuchunshuo@huawei.com >
2025-11-09 04:45:27 +09:00
Jaya Yuan and GitHub
ea97940d6c
[DCP] Support Decode Context Parallel (DCP) for GQA with FlashAttention ( #24864 )
...
Signed-off-by: yuanyongjie.yyj <yuanyongjie.yyj@antgroup.com >
Signed-off-by: FENP <32334296+FENP@users.noreply.github.com >
Signed-off-by: Jaya Yuan <yuanyongjie.yyj@antgroup.com >
2025-10-14 13:07:50 +00:00
Harry Mellor and GitHub
8fcaaf6a16
Update Optional[x] -> x | None and Union[x, y] to x | y ( #26633 )
...
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com >
2025-10-12 09:51:31 -07:00
Harry Mellor and GitHub
2f99f2f506
Tidy vllm/config/__init__.py to only add classes and functions ( #26405 )
...
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com >
2025-10-08 07:10:00 -07:00
Cyrus Leung and GitHub
1e4ecca1d0
[V0 Deprecation] Remove VLLM_USE_V1 from tests ( #26341 )
...
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk >
2025-10-07 15:42:31 +00:00
Harry Mellor and GitHub
d6953beb91
Convert formatting to use ruff instead of yapf + isort ( #26247 )
...
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com >
2025-10-05 07:06:22 -07:00
Ming Yang and GitHub
4e5affeaa1
[CI] Add Decode Context Parallelism (DCP) test to CI ( #24487 )
...
Signed-off-by: Ming Yang <minos.future@gmail.com >
2025-09-16 21:21:28 +08:00
ac201a0eaf
[Feature] Support Decode Context Parallel (DCP) for MLA ( #23734 )
...
Signed-off-by: hongchao <hongchao@msh.team >
Signed-off-by: youkaichao <youkaichao@gmail.com >
Co-authored-by: hongchao <hongchao@msh.team >
Co-authored-by: youkaichao <youkaichao@gmail.com >
2025-09-06 13:24:05 +08:00
