EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity
5,105 Commits 8 Branches 0 Tags
a5a03209e9598c25f28adb29a70c4ab6dc205e61
Commit Graph

3439 Commits

Author SHA1 Message Date
Cheng Wan
a5a03209e9 Fix circular import (#10107) 2025-09-06 01:34:17 -07:00
Cheng Wan
21af5c0404 [Fix] Compatibility between DP attention and pipeline parallelism (#10100) 
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
 2025-09-06 01:34:10 -07:00
Jinyang Yuan
012584ecd5 perf: Avoid unnecessary data type conversions for DeepSeek-V3 on Blackwell (#9834) 
Signed-off-by: Jinyang Yuan <154768711+jinyangyuan-nvidia@users.noreply.github.com>
 2025-09-06 14:06:46 +08:00
Kaixi Hou
90dfe3de4c [NVIDIA] disable chunked prefix cache when dp and blackwell is used (#9861) 2025-09-06 14:05:16 +08:00
Kaixi Hou
9a719b7afc [NVIDIA] Remove unused get_fused_moe_impl_class function (#9764) 2025-09-05 22:41:22 -07:00
Cheng Wan
3fa62da78c [7/N] MoE Refactor: the implementation of new framework (#9269) 2025-09-05 21:09:09 -07:00
Chi-Chih Chang
ad26f298e2 fix double sparsity initialization (#6905) 2025-09-06 11:45:24 +08:00
sogalin
8d114f254b Fix RMSNorm API CALL mismatch issue. (#10032) 
Co-authored-by: Hubert Lu <Hubert.Lu@amd.com>
 2025-09-05 20:45:13 -07:00
Zhiqiang Xie
0b8c5721f1 [HiStorage] Remove delete and clear as necessary methods (#10039) 2025-09-06 10:27:26 +08:00
gongwei-130
ab62b135c1 support Llama4 with non uniformed intermediate size across layers for… (#10047) 2025-09-05 17:28:15 -07:00
Xinyuan Tong
273b28344b [Minor] Refactors KV memory pool (#9842) 2025-09-05 17:06:08 -07:00
pansicheng
f84db115b1 Add storage read/write bandwidth logs to monitor kvcache performance (#9965) 
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
 2025-09-05 16:52:55 -07:00
jacky.cheng
efb0de2c8d Update wave-lang to 3.7.0 and unify Wave kernel buffer options (#10069) 2025-09-05 16:01:52 -07:00
Adam Yanxiao Zhao
0f6ac5e21d [Bug Fix] Fix Glm4vVisionBlock norm (#9884) 2025-09-05 14:20:36 -07:00
hlu1
2985090084 Update flashinfer to 0.3.1 for B300 support (#10087) 
Signed-off-by: Hao Lu <14827759+hlu1@users.noreply.github.com>
 2025-09-05 13:41:01 -07:00
Mahmoud Ashraf
e678cc717d [bugfix]: use correct cache location for cross attention in torch native backend (#8622) 2025-09-05 13:39:46 -07:00
Morpheus Guo
4efe844a25 enable aiter gemm_a8w8_bpreshuffle for ptpc gemm (#8555) 2025-09-05 12:54:40 -07:00
fzyzcjy
5e5c30d9ab Tiny let DeepGEMM scale checks cover more cases (#7182) 
Co-authored-by: Yineng Zhang <me@zhyncs.com>
 2025-09-05 19:52:32 +08:00
fzyzcjy
9f00ec44eb Fix and enhance dumper (#8725) 2025-09-05 19:51:09 +08:00
fzyzcjy
8e85ee887e Support simple evals in text comparator (#8867) 2025-09-05 19:50:21 +08:00
fzyzcjy
adf73175d6 Forbid DeepEP racing condition when too many tokens (#9567) 2025-09-05 19:47:05 +08:00
DevashishLal-CB
13705dae06 [Fix] Add speculative_draft_model_revision to server_args (#5255) 
Signed-off-by: Devashish Lal <devashish@rivosinc.com>
 2025-09-05 19:45:46 +08:00
fzyzcjy
df97b31f37 Tiny support setting numa nodes for different ranks (#10006) 2025-09-05 19:01:27 +08:00
fzyzcjy
339f8eef09 [1/2] Optimizations and refactors about quant kernel (#9534) 2025-09-05 18:45:08 +08:00
limingshu
afd9f2f560 Fix typo in scheduler (#9934) 2025-09-05 17:45:27 +08:00
Jimmy
f40038fb09 [Vulnerability]feat(conn): set bootstrap server host (#9931) 2025-09-05 17:36:17 +08:00
Elfie Guo
bebd0576e5 Integrate trtllm ragged attention for prefill self-attention (#9801) 2025-09-05 17:18:00 +08:00
Huang Long
f98366604b fix MultiTokenizerWrapper name (#10049) 
Signed-off-by: huanglong <huanglong@linux.alibaba.com>
 2025-09-05 13:39:46 +08:00
Liangsheng Yin
6e95f5e5bd Simplify Router arguments passing and build it in docker image (#9964) 2025-09-05 12:13:55 +08:00
Yineng Zhang
fa9c82d339 chore: bump v0.5.2rc2 (#10050) 2025-09-04 20:07:27 -07:00
kk
918e3d4c27 Fix accuracy drop of dsv3 run in dp enablement (#8677) 
Co-authored-by: wunhuang <wunhuang@amd.com>
 2025-09-04 16:51:16 -07:00
kk
e96973742c Optimized deepseek-v3/r1 model performance on mxfp4 run (#10008) 
Co-authored-by: wunhuang <wunhuang@amd.com>
Co-authored-by: HAI <hixiao@gmail.com>
Co-authored-by: Hubert Lu <55214931+hubertlu-tw@users.noreply.github.com>
 2025-09-04 15:11:22 -07:00
ykwd
93088b6975 [Hicache] Mooncake API Fix & Test, and Improved Readme (#9951) 
Co-authored-by: Teng Ma <sima.mt@alibaba-inc.com>
 2025-09-04 13:55:39 -07:00
Cheng Wan
453511acc7 Save memory for expert model parallel (#9957) 2025-09-04 13:31:47 -07:00
pansicheng
d07304870b fix 3fs zerocopy (#9938) 
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
 2025-09-04 13:24:12 -07:00
Yingchun Lai
b32ab0705e metrics: support customer buckets for prompt/generation_tokens_histogram (#9634) 2025-09-04 22:22:08 +08:00
Huapeng Zhou
75ee00112d [Doc] Fix SGLang tool parser doc (#9886) 2025-09-04 21:52:53 +08:00
Yuan Luo
ec15c8360e Optimize Qwen3-moe model by using flashinfer fused allreduce (#9973) 
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
 2025-09-04 20:48:53 +08:00
hzh0425
106c2b31fb feat(hicache): Add generic hicache ci e2e test and benchmark test (#9846) 
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
 2025-09-04 20:43:46 +08:00
pansicheng
c67569491c Ensure chunked request extension length respects both rem_chunk_tokens and rem_total_tokens limits (#10003) 2025-09-04 04:15:26 -07:00
Liangsheng Yin
27e8ffed37 [1/N] DP-refactor: move dp balance code into scheduler's mixin class (#10004) 2025-09-04 16:53:58 +08:00
wxsm
4dbb34fe43 fix: health_generate endpoint in mini_lb (#9997) 2025-09-04 01:52:28 -07:00
Tony Lu
1e18a341e9 [Bugfix] fix pd chat completion protocol for batching support (#10016) 
Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
 2025-09-04 01:43:16 -07:00
Hubert Lu
2c562fd2d0 Fix Llama 4 with MXFP4 dynamic quant on MI35x (#9993) 2025-09-04 00:48:58 -07:00
Lianmin Zheng
397448ebbc [Auto Sync] Update parallel_state.py, few_shot_gsm8k.py (20250903) (#9986) 
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Leon Gao <leon.gao19@gmail.com>
 2025-09-03 16:55:43 -07:00
Elfie Guo
66d5d0425c Minor update regarding issue #9704 (#9733) 2025-09-03 16:52:07 -07:00
Szymon Ożóg
8cbf71dc2d Triton 3.4.0 MoE config for Deepseek TP16 H100 (#9978) 2025-09-03 13:16:16 -07:00
Xinyuan Tong
56eb5d0a3d fix swa clear(): rename is_in_free_group to is_not_in_free_group (#9914) 2025-09-03 11:42:12 -07:00
timmy-feng
4ed9053ecf Remove mrope position sync (#9460) 
Co-authored-by: Nathan Wang <nathan.r.wang@gmail.com>
 2025-09-03 11:40:53 -07:00
Yineng Zhang
1b2ff4fb7f Revert "Optimized deepseek-v3/r1 model performance on mxfp4 run (#9671)" (#9959) 2025-09-03 00:50:04 -07:00