Commits · 3d6be1fbce6c36dc090cdc8628f0ab88a0003b5a · change / sglang

14 Aug, 2025 15 commits
- add w8a8-fp8-block-wise H20-3e triton config (#8018) · 3d6be1fb
  lukec authored Aug 14, 2025
  
  3d6be1fb
- Add H200 fused MoE kernel configs for DeepSeek-V3 in triton 3.3.1 (#7687) · 4063234c
  Jun Liu authored Aug 14, 2025
  
  4063234c
- Add H20 fused MoE kernel configs for Dpsk & Qwen3 (#7631) · 83feef5b
  Tommy Yang authored Aug 14, 2025
  
  83feef5b
- Add Triton Fused MoE kernel config for E=16 on B200 (#7004) · 2871eacc
  Brayden Zhong authored Aug 14, 2025
  
  2871eacc
- Add H200 fused MoE kernel tuning configs for Qwen3-Coder-480B-A35B-Instruct (#8852) · ac15bdc1
  forestlee95 authored Aug 14, 2025
  
  ac15bdc1
- Add A800 fused MoE kernel tuning configs for GLM4.5 and GLM4.5-Air (#8808) · d6451c3f
  Li Hui authored Aug 14, 2025
  
  d6451c3f
- [Perf] Tunings for SM100 FP8 CUTLASS kernel (#8818) · 841810f2
  henryg authored Aug 13, 2025
  
  841810f2
- fix io group (#9154) · 733446dd
  pansicheng authored Aug 14, 2025
```
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
```
  733446dd
- Feature: support qwen and llama4 reducescatter for dp attention padding (#9101) · 4c22897a
  wxzhoucs authored Aug 14, 2025
  
  4c22897a
- Faster weight processing (trtllm-gen moe nvfp4) (#9162) · 1bc183c6
  Alex Yang authored Aug 13, 2025
  
  1bc183c6
- [DP Attention] Refactor: adding some utility functions (#9136) · b87aacb5
  Cheng Wan authored Aug 13, 2025
  
  b87aacb5
- Fix docker container DeepEP error on Blackwell (#9171) · b3363cc1
  fzyzcjy authored Aug 14, 2025
  
  b3363cc1
- [Bugfix] Avoid unnecessary reduce-scatter call in prepare_mlp (#9169) · 98457c04
  Huaixin Chang authored Aug 14, 2025
  
  98457c04
- [AMD] Update fallback images for AMD CI (#9159) · 0fc8bf2c
  michael-amd authored Aug 13, 2025
```
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  0fc8bf2c
- Replace `sglang.srt.layers.quantization.scalar_types` with `sgl_kernel.scalar_type` (#8951) · a669bc2f
  Hongbo Xu authored Aug 14, 2025
  
  a669bc2f
13 Aug, 2025 19 commits
- Fix broken trtllm_mha attn backend with gpt-oss (#9161) · 6b7c2471
  Nicolas Castet authored Aug 13, 2025
  
  6b7c2471
- [Generative Score API] Optimization to Remove Decode. (#8840) · a027a9b4
  Sundara Raman Ramachandran authored Aug 13, 2025
  
  a027a9b4
- Clean up allocators (#9134) · 9e426466
  Lianmin Zheng authored Aug 13, 2025
```
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  9e426466
- Swap xeon ci to gnr server (#9042) · 2f20f430
  DiweiSun authored Aug 14, 2025
  
  2f20f430
- [Model] Support Qwen3ForSequenceClassification for Qwen3-Embed Model (#7957) · 65736dc5
  Zhihao Liu authored Aug 14, 2025
  
  65736dc5
- chore: bump v0.5.0rc1 (#9069) · 7b56e494
  Yineng Zhang authored Aug 13, 2025
  
  7b56e494
- Support FA3 backend for gpt-oss (#9028) · 0ff6d1fc
  Ke Bao authored Aug 14, 2025
  
  0ff6d1fc
- [PD] feat: mooncake use batch reg/dereg (#8910) · 4a16a71c
  Teng Ma authored Aug 14, 2025
```
Co-authored-by: Shangming Cai <csmthu@gmail.com>
```
  4a16a71c
- [PD] optimize kv cache transfer directly using batch transfer (#9149) · a16923ef
  Francis authored Aug 14, 2025
```
Co-authored-by: Shangming Cai <csmthu@gmail.com>
```
  a16923ef
- [router] optimize Rust compilation and development workflow (#9133) · 6337d905
  Simo Lin authored Aug 13, 2025
  
  6337d905
- feat: update fa3 (#9126) · 71fb8c95
  Yineng Zhang authored Aug 13, 2025
  
  71fb8c95
- Update fa3 interface and add unit test (#9150) · 94f44b88
  Ke Bao authored Aug 13, 2025
  
  94f44b88
- Double vision prefill throughput by defaulting to optimal vision attention backend (#8484) · 3b3b3baf
  Kevin Xiang Li authored Aug 13, 2025
```
Co-authored-by: Xiang (Kevin) Li <lik@nvidia.com>
```
  3b3b3baf
- Update docker file for MI35x base image update to support gpt-oss mxfp4 model (#9111) · 35e6bc92
  kk authored Aug 13, 2025
```
Co-authored-by: wunhuang <wunhuang@amd.com>
```
  35e6bc92
- Fix gpt-oss ~2x memory consumption issue (#9146) · 9394ed63
  fzyzcjy authored Aug 13, 2025
  
  9394ed63
- Support Triton FP8 Gemm can handle hidden_dim not divisible by 16 (#9093) · 930fe467
  Stefan He authored Aug 12, 2025
```
Co-authored-by: Xiaoyu Zhang <35585791+BBuf@users.noreply.github.com>
```
  930fe467
- [1/2][resubmit again] sgl-kernel: Fuse routed scaling factor into moe_fused_gate (#9088) · 13c48dcf
  Trevor Morris authored Aug 12, 2025
  
  13c48dcf
- Use FlashInfer's TRTLLM FP8 Blockscale GEMM (#8588) · 8723b4f1
  Elfie Guo authored Aug 12, 2025
  
  8723b4f1
- fix: wrong docker hub org name (#9137) · 62f99e08
  li chaoran authored Aug 13, 2025
```
Signed-off-by: mywaaagh_admin <pkwarcraft@gmail.com>
```
  62f99e08
12 Aug, 2025 6 commits

[Feature] Support custom set kv buffer kernel (#8884) · 86a0be65
DarkSharpness authored Aug 12, 2025

86a0be65
Support page first layout zero copy for mooncake store (#8651) · 0edda320
huangtingwei authored Aug 13, 2025
```
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
```
0edda320
chore: use cp310 (#9130) · 924827c3
Yineng Zhang authored Aug 12, 2025

924827c3
fix: update Dockerfile (#9129) · c81daf83
Yineng Zhang authored Aug 12, 2025

c81daf83

[AMD] Support Wave attention backend with AMD GPU optimizations (#8660) · 25caa7a8

jacky.cheng authored Aug 13, 2025


Signed-off-by: Stanley Winata <stanley.winata@amd.com>
Signed-off-by: Harsh Menon <harsh@nod-labs.com>
Signed-off-by: nithinsubbiah <nithinsubbiah@gmail.com>
Signed-off-by: Ivan Butygin <ivan.butygin@gmail.com>
Signed-off-by: xintin <gaurav.verma@amd.com>
Co-authored-by: Harsh Menon <harsh@nod-labs.com>
Co-authored-by: Stanley Winata <stanley.winata@amd.com>
Co-authored-by: Stanley Winata <68087699+raikonenfnu@users.noreply.github.com>
Co-authored-by: Stanley Winata <stanley@nod-labs.com>
Co-authored-by: Ivan Butygin <ivan.butygin@gmail.com>
Co-authored-by: nithinsubbiah <nithinsubbiah@gmail.com>
Co-authored-by: Nithin Meganathan <18070964+nithinsubbiah@users.noreply.github.com>
Co-authored-by: Ivan Butygin <ibutygin@amd.com>

25caa7a8

Fix typos in supported models documentation (#9119) · 03d11449
Hangzhi authored Aug 12, 2025

03d11449