Commit Graph
69 Commits
Author SHA1 Message Date
Shangming Cai eccee4c48e [PD] Change bootstrap_room metadata dtype from int64 to uint64 (#19141) 2026-02-22 14:20:16 +08:00
Shangming Cai 1f39bf6523 [Bugfix] Add warnings when NSA indexer cache indice mismatch in PD module (#18727) 2026-02-13 15:20:05 +08:00
Shangming Cai 2a8a48c0ca Reuse initialized transfer engine in mooncake store (#18460)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-02-13 01:21:35 +08:00
Shangming Cai bffd765417 Refactoring Mooncake TE as a shared distributed component (#17810)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-02-09 10:53:11 +08:00
Shangming Cai 52401bec1d chore: bump mooncake version to 0.3.9 (#18316)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-02-07 17:30:01 +08:00
Shangming Cai afae4c7178 [PD] Minor code cleanup for mooncake backend (#18279)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-02-05 17:38:09 +08:00
Shangming Cai 3ad3268e06 [CI] Skip PD hybrid attention test with different TP temporarily (#17791) 2026-01-27 10:59:33 +08:00
Shangming Cai 2262c5c9b5 Add ZhengWG to CI_Permission (#17572) 2026-01-22 17:24:06 +08:00
Shangming Cai 23d765d1f2 [Doc] Update pipeline parallelism documentation (#17414) 2026-01-20 20:10:56 +08:00
Shangming Cai 71cb9d0302 [PD] Fix PP dynamic chunking with DP attention (#17339)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-01-20 11:27:13 +08:00
Shangming Cai 0227db8926 [PD] Optimize MHA models pp util calculation logic (#17306) 2026-01-19 11:23:15 +08:00
Shangming Cai 4c59782e0f Fix hybrid attention PD Disaggregation test (#17099)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-01-15 23:38:58 +08:00
Shangming Cai 4b14f622e1 [CI] Add PD Disaggregation aarch64 test (#16572)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-01-10 14:44:54 +08:00
Shangming Cai 0c4e155a3c chore: bump mooncake version to 0.3.8.post1 (#16792)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-01-09 18:42:27 +08:00
Shangming Cai 973116e6bb [Doc] Optimize pipeline parallelism doc (#16630)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-01-07 14:52:42 +08:00
Shangming Cai 130f60ee78 Add CODEOWNERS for scheduler_pp_mixin.py (#16481) 2026-01-05 21:12:20 +08:00
Shangming Cai 9cac3c8628 Add liusy58 into CI_PERMISSION (#16373)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-01-03 21:36:35 -08:00
Shangming Cai db3821a9ef [PP] Add a minimum chunk value for PP dynamic chunking (#16140)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-30 16:35:39 +08:00
Shangming Cai 41addd2e08 chore: bump mooncake version to 0.3.8 (#15886)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-27 23:34:24 +08:00
Shangming Cai b3f83cc1c5 Fix pipeline parallelism doc typos (#15688)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-23 22:19:15 +08:00
Shangming Cai c15fa1c577 [2/N] Update doc of Pipeline Parallelism with case study (#15684)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-23 20:56:15 +08:00
Shangming Cai 17e81c750a [PP] Add dynamic chunking PP test (#15395)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-19 01:09:40 +08:00
Shangming Cai ee1ca51de8 [PP] Fix dynamic chunking strategy for PP (#15372)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-18 14:24:55 +08:00
Shangming Cai eeb2b9b259 [PP] Minor code cleanup for Pipeline Parallelism (#15329)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-17 23:31:25 +08:00
Shangming Cai 6682475124 [CI] Improve flaky 4 GPU test success rate (#15234) 2025-12-17 00:02:26 +08:00
Shangming Caiandybyang 36fcf71fff [Qwen3-next] Add PD disaggregation support for mamba with extra_buffer (#15180)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
Co-authored-by: ybyang <10629930+whybeyoung@users.noreply.github.com>
2025-12-16 14:36:00 +08:00
Shangming Cai d277a86dea [CI] Add disaggregation decode PP test (#15114) 2025-12-14 23:54:25 +08:00
Shangming Caiandalpha-baby c8cf1cafdb [1/N] Update doc of Pipeline Parallelism (#14985)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
Co-authored-by: alpha-baby <fujianhao1997@qq.com>
2025-12-12 19:32:52 +08:00
Shangming Cai a076d75e98 [Fix] Remove unused import from test_disaggregation_hicache.py (#14880) 2025-12-11 15:50:47 +08:00
Shangming Cai 93452a8252 [PD] Support decode pp for PD disaggregation (#14265)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-12-03 14:35:29 +08:00
Shangming Cai c6d34a0688 Move piecewise cuda graph test to manual dir to fix CI (#14121)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-11-29 00:52:01 -08:00
Shangming Cai b704b0a922 Optimize uneven PP layer distribution logic to improve PP performance (#13977) 2025-11-26 22:05:03 +08:00
Shangming Cai 4c3573e493 [PD] Clarify init method docstrings for kvsender and kvreceiver (#13476) 2025-11-18 12:20:19 +08:00
Shangming Cai c2d4716da0 chore: bump mooncake version to 0.3.7.post2 (#12599)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-11-04 01:09:31 -08:00
Shangming Cai ab8b83f71d chore: upgrade mooncake 0.3.7.post1 (#12541)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-11-03 00:34:32 -08:00
Shangming Cai 358ae3563d Tiny fix eos handling for PD disaggregation (#12334)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-31 17:57:10 -07:00
Shangming Cai 41a113356a Fix potential eos bug on decode instance when PD is enabled (#12206)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-28 01:29:02 -07:00
Shangming Cai 7fceeef599 Fix flaky hicache test with mooncake backend (#11953)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-22 21:00:47 +08:00
Shangming Cai f3cd5d2510 [CI] Fix b200 flashinfer installation (#11915)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-21 22:28:50 +08:00
Shangming Cai 05d3667ab9 [CI] disable glm4.1v and fix the flashinfer installation (#11902)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-21 18:38:35 +08:00
Shangming Cai a2ba0bc3df Tiny clean up for PD module and doc (#11747)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-20 11:52:42 +08:00
Shangming Cai 1de3924b18 [CI] Add GLM4MoE model test (#11706)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-16 16:25:58 +08:00
Shangming Cai 476c67d7fc Fix missing a2a backend init of GLM4.5 MoE Block (#11692)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-15 19:13:08 -07:00
868403f642 [PD] Add PD support for hybrid model (Qwen3-Next, DeepSeek V3.2 Exp) (#10912)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
Co-authored-by: hzh0425 <hzh0425@apache.org>
Co-authored-by: ZeldaHuang <hzm414167@alibaba-inc.com>
2025-10-15 18:59:14 -07:00
Shangming CaiandLiangsheng Yin 6d0364681c Fix 1-step draft model forward (#11653)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
Co-authored-by: Liangsheng Yin <lsyincs@gmail.com>
2025-10-15 19:11:33 +08:00
Shangming Cai c5fe3c0b75 Tiny fix test run estimated time (#11544)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-13 02:23:13 -07:00
Shangming Cai 70fbb3adf6 [CI] Refactor PD disaggregation test suite (#11363)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-09 18:50:39 -07:00
Shangming Cai 0a7c4bded7 [Doc] Update mooncake nvlink transport doc for PD disaggregation (#11321)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-08 00:59:29 -07:00
Shangming Cai c560410da7 Refactor and optimize mooncake CI (#11162)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-05 14:08:52 -07:00
Shangming Cai 2c7f4ca2f2 Optimize debug log position of PD abort request (#11090)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-03 23:07:02 +08:00
Shangming Cai 9c339d6b47 [PD] Extract the PP transfer layer calculate logic from Mooncake to Common backend (#10565)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-28 00:10:41 +08:00
Shangming Cai e23e280e16 Add support for topk metadata transferring for PD (#10616)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-28 00:09:38 +08:00
Shangming Cai 23632d350c Fix latest main ci (#10799)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-23 12:46:13 -07:00
Shangming Cai d21c35224d Fix hicache mooncake backend CI (#10792)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-23 02:04:44 -07:00
Shangming Cai 70e4b21853 Fix flaky logprobs test (#10728)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-22 00:46:26 -07:00
Shangming Cai 74cd6e3902 chore: upgrade mooncake 0.3.6.post1 to fix gb200 dockerfile (#10681)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-20 00:12:26 -07:00
Shangming Cai 60fc5b51f6 chore: upgrade mooncake 0.3.6 (#10596)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-18 00:19:30 -07:00
Shangming Cai a13dd1e492 [PD] Improve disaggregation common backend and refactor mooncake backend (#10273)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-17 22:58:07 -07:00
Shangming Cai edd6a07bc0 Minor fix lint introduced by #10466 (#10507)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-16 01:38:25 -07:00
Shangming Cai f5f6b3b4b5 Refactor fused_add_rmsnorm import logic (#10207)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-09 00:23:58 -07:00
Shangming Cai 00974e4f6e [CI] Refactor disaggregation tests (#10068)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-06 22:14:46 +08:00
Shangming Cai a25e8e42eb Move multi-tokenizer event loop to better place (#9902)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-09-01 23:12:21 -07:00
Shangming Cai 25ef53f05f [PD] Fix nvlink transport accuracy through transferring metadata with tcp (#9261)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-08-20 19:29:10 -07:00
Shangming Cai d8ed60f254 [CI] Fix disaggregation failure tolerance CI (#9378)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-08-19 23:31:08 -07:00
384f8ab5ce [PD] Support PD disaggregation with Prefill PP (#8846)
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
Signed-off-by: Shangming Cai <csmthu@gmail.com>
Co-authored-by: root <huzhiyuan@xiaohongshu.com>
Co-authored-by: Ying Sheng <sqy1415@gmail.com>
Co-authored-by: Francis <38564764+ssssnow@users.noreply.github.com>
Co-authored-by: zitto <zhjc1124@gmail.com>
2025-08-16 18:31:31 -07:00
Shangming Cai 8ca07bd948 [CI] Fix sgl-router disaggregation test (#9222)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-08-15 02:24:44 -07:00
Shangming Cai 4caca4f6b4 Fix typo in REVIEWERS (#9113)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-08-12 11:55:49 -07:00
Shangming Cai eb429b88a4 [PD] Respect sampling_params.max_new_tokens when PD disaggregation is activated (#7598)
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
2025-06-27 22:22:01 -07:00
shangmingc 89a554181f [PD] Fix unclosed prefill connection warning of mini_lb (#5155)
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
2025-04-08 09:15:06 -07:00