Commits · 4a0d19198bf9222edcb9879028990b481f8ffe56 · change / sglang · GitLab

10 Jul, 2025 1 commit
- Fix bug of deepseek-v3 under DP+EP mode with large batchsize/seqlen (#6449) · 4a0d1919
  likesen-alibaba authored Jul 10, 2025
  
  4a0d1919
27 Jun, 2025 1 commit
- Move multimodal processors into a separate folder (#7581) · ce3a3e87
  Lianmin Zheng authored Jun 27, 2025
  
  ce3a3e87
16 Jun, 2025 1 commit
- [fix] fix DeepGEMM blackwell input quant & ut & fix style and log (#7247) · 5ca07eed
  JieXin Liang authored Jun 17, 2025
  
  5ca07eed
14 Jun, 2025 2 commits
- Support new DeepGEMM (#7172) · 93cec433
  fzyzcjy authored Jun 14, 2025
  
  93cec433
- Refactor DeepGEMM integration (#7150) · b4c41f72
  fzyzcjy authored Jun 14, 2025
  
  b4c41f72
13 Jun, 2025 1 commit
- Support new DeepGEMM format in per token group quant (part 2: srt) (#7155) · bec3e484
  fzyzcjy authored Jun 14, 2025
  
  bec3e484
29 May, 2025 1 commit
- refactor apply_w8a8_block_fp8_linear in fp (#6545) · 485a023b
  ChangyiYang authored May 29, 2025
  
  485a023b
11 May, 2025 1 commit
- Add typo checker in pre-commit (#6179) · 2ce87935
  applesaucethebun authored May 11, 2025
```
Co-authored-by: Brayden Zhong <b8zhong@uwaterloo.ca>
```
  2ce87935
08 May, 2025 3 commits
- DeepEP normal support deepgemm-contiguous (#5626) · acc816d8
  lukec authored May 08, 2025
```
Co-authored-by: Yingyi Huang <yingyihuang2000@outlook.com>
Co-authored-by: Cheng Wan <54331508+ch-wan@users.noreply.github.com>
Co-authored-by: Xuting Zhou <xutingz@nvidia.com>
Co-authored-by: ZhengHSI <zhenghsi@qq.com>
```
  acc816d8
- Clean logs for DeepSeek-V3 launching (#6079) · 73600673
  Baizhou Zhang authored May 07, 2025
  
  73600673
- [refactor] slightly tidy fp8 module (#5993) · b70957fc
  JieXin Liang authored May 08, 2025
  
  b70957fc
21 Apr, 2025 1 commit
- [feature] enable pre compile jit deep_gemm (#5580) · c2942907
  JieXin Liang authored Apr 22, 2025
  
  c2942907
20 Apr, 2025 1 commit
- [perf] introduce deep gemm group_gemm_masked as bmm (#5432) · 99456bca
  JieXin Liang authored Apr 20, 2025
  
  99456bca
19 Apr, 2025 1 commit
- Remove one kernel in per_tensor_quant_mla_fp8 (#5549) · 613b197e
  fzyzcjy authored Apr 20, 2025
  
  613b197e
18 Apr, 2025 1 commit
- use sglang_per_token_group_quant_fp8 from sgl-kernel instead of trion kernel (#5473) · b2a189dd
  strgrb authored Apr 18, 2025
```
Co-authored-by: Zhang Kaihong <zhangkaihong.zkh@alibaba-inc.com>
```
  b2a189dd
16 Apr, 2025 1 commit
- Clean up imports (#5467) · 177320a5
  Lianmin Zheng authored Apr 16, 2025
  
  177320a5
15 Apr, 2025 1 commit
- [Fix] Turn off DeepGEMM by default (#5263) · f6772f14
  Baizhou Zhang authored Apr 14, 2025
  
  f6772f14
14 Apr, 2025 1 commit
- [perf] experimental enhance fp8 per-tensor quant (#5370) · bdde2375
  JieXin Liang authored Apr 15, 2025
  
  bdde2375
13 Apr, 2025 1 commit
- chore: upgrade sgl-kernel 0.0.8.post3 (#5342) · f58b929a
  Yineng Zhang authored Apr 13, 2025
  
  f58b929a
12 Apr, 2025 1 commit
- fix: use deepgemm only on hopper (#5310) · 61172091
  Yineng Zhang authored Apr 11, 2025
  
  61172091
09 Apr, 2025 1 commit
- feat: add DeepGEMM build warning (#5176) · 6669d127
  Yineng Zhang authored Apr 08, 2025
```
Co-authored-by: grimoire <streetyao@live.com>
```
  6669d127
28 Mar, 2025 1 commit
- Clean up `import vllm` in quantization/__init__.py (#4834) · 74e0ac1d
  Lianmin Zheng authored Mar 28, 2025
  
  74e0ac1d
21 Mar, 2025 1 commit
- Set deepgemm to the default value in the hopper architecture. (#4613) · 4c56e5db
  lukec authored Mar 21, 2025
  
  4c56e5db
19 Mar, 2025 1 commit
- Create col-major and tma-aligned x_scale for deep_gemm.gemm_fp8_fp8_bf16_nt (#4515) · f9c53cbb
  strgrb authored Mar 19, 2025
```
Co-authored-by: Zhang Kaihong <zhangkaihong.zkh@alibaba-inc.com>
```
  f9c53cbb
13 Mar, 2025 4 commits
- Revert "[XPU][CPU] Enable the native path of DeepSeek" (#4367) · 45de8971
  Lianmin Zheng authored Mar 12, 2025
  
  45de8971
- [XPU][CPU] Enable the native path of DeepSeek (#4086) · 71046fcd
  Meng, Hengyu authored Mar 13, 2025
```
Co-authored-by: Zhang, Liangang <liangang.zhang@intel.com>
```
  71046fcd
- Support page size > 1 (#4356) · c76040e3
  Lianmin Zheng authored Mar 12, 2025
  
  c76040e3
- [FIX] fix incorrect output when enable both deepgemm and torch compile (#4359) · 85ef7f64
  AniZpZ authored Mar 13, 2025
```
Co-authored-by: xuyongfei.xyf <xuyongfei.xyf@antgroup.com>
```
  85ef7f64
12 Mar, 2025 1 commit
- unify is_cuda and is_hip (#4321) · d1da58e2
  Yineng Zhang authored Mar 11, 2025
  
  d1da58e2
11 Mar, 2025 1 commit
- linear support deepgemm (#4199) · dce303e2
  lukec authored Mar 11, 2025
```
Co-authored-by: yinfan98 <1106310035@qq.com>
```
  dce303e2
09 Mar, 2025 1 commit
- Apply sgl w8a8 fp8 kernel (#3148) · 0dd6cda2
  HandH1998 authored Mar 09, 2025
  
  0dd6cda2
17 Feb, 2025 1 commit
- Revert "[ROCm] Use `tl.range()` in block GEMM kernels with `num_stage… (#3632) · 5f1a485d
  Yineng Zhang authored Feb 17, 2025
  
  5f1a485d
16 Feb, 2025 1 commit
- [ROCm] Use `tl.range()` in block GEMM kernels with `num_stages` set by host. (#3535) · 03caefeb
  Wen-Heng (Jack) Chung authored Feb 16, 2025
```
Co-authored-by: HAI <hixiao@gmail.com>
```
  03caefeb
12 Feb, 2025 2 commits
- integrate blockwise fp8 kernel (#3529) · 98eecbda
  yizhang2077 authored Feb 13, 2025
  
  98eecbda
- use sgl_per_token_group_quant_fp8 kernel (#3493) · 45e3a7bc
  Xiaoyu Zhang authored Feb 12, 2025
  
  45e3a7bc
06 Feb, 2025 1 commit
- [ROCm] Fix fp8 unrolledx4 matmul kernel. (#3325) · 32de54ed
  Wen-Heng (Jack) Chung authored Feb 05, 2025
```
Co-authored-by: HAI <hixiao@gmail.com>
```
  32de54ed
05 Feb, 2025 1 commit
- [ROCm] Logic to decide whether to used manually unrolled kernel. (#3306) · 7ab84948
  Wen-Heng (Jack) Chung authored Feb 04, 2025
  
  7ab84948
04 Feb, 2025 1 commit
- [ROCm] Manually unroll _w8a8_block_fp8_matmul kernel on AMD GPU. (#3299) · c2723a42
  Wen-Heng (Jack) Chung authored Feb 04, 2025
  
  c2723a42
29 Dec, 2024 2 commits
- Update Triton configs for block fp8 kernels (#2641) · afa0341e
  HandH1998 authored Dec 29, 2024
  
  afa0341e
- AMD: set weights and scaling numbers properly for block FP8 (#2637) · 30828e71
  HAI authored Dec 29, 2024
  
  30828e71