Add KV4-capable backend flashmla and update server args#14989
Conversation
- Added the new KV4-capable backend flashmla and documented it in attention_backend.md - Updated server_args.py accordingly Signed-off-by: Ho-Ren (Jack) Chuang <horenchuang@bytedance.com>
Summary of ChangesHello @JackChuang, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request expands the attention backend capabilities by integrating Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
There was a problem hiding this comment.
Code Review
This pull request correctly adds support for the flashmla attention backend with FP4 KV cache, including documentation and server argument updates. The changes are well-aligned with the PR's goal. I have one minor suggestion to improve code maintainability by sorting the list of backend choices.
I noticed that the new file for the flashmla backend and its registration were not in the provided diff, so I couldn't perform a line-by-line review. However, from a high-level look, the implementation seems consistent with existing backends.
It's also worth noting that the checklist indicates unit tests are missing. Adding tests for the new backend would be valuable for ensuring long-term stability and correctness.
|
/tag-and-rerun-ci |
…n_eagle3_npu * 'main' of https://github.com/sgl-project/sglang: (121 commits) Super tiny add gsp-fast-prepare (sgl-project#14992) Super tiny fix confusing slash_command_handler hint (sgl-project#14976) Super tiny remove unused argument (sgl-project#14966) [registry] Add a strict mode to model registration (sgl-project#14933) Feature/Fix multi lora scheduler blocking issue and evict LoRA None lastly (sgl-project#14795) Tune triton fused moe for the case of glm-4.6-fp8 b200 tp4 (sgl-project#15020) [model-gateway] refactor: unify worker management into modular workflow structure (sgl-project#15010) Update ci permission (sgl-project#15014) Refactor of http and engine entrypoints to allow custom override (sgl-project#14869) Add KV4-capable backend flashmla and update server args (sgl-project#14989) Revert several PRs (sgl-project#14958) Super tiny extract route_typed_request_once (sgl-project#14951) Fix CI by reverting incorrect metric check logic (sgl-project#15004) [model-gateway] refactor: workflow engine cleanup and minor optimization (sgl-project#15001) [model-gateway] fix: handle workflow deadlock and optimize cycle detection (sgl-project#15000) [model-gateway] feat: add DAG parallel execution support and workflow optimization (sgl-project#14999) [model-gateway] refactor: extract workflow engine to src/workflow module (sgl-project#14996) Update CODEOWNERS for multimodal_gen (sgl-project#14995) [diffusion] docker: Tiny fix Docker Hub link in installation documentation (sgl-project#14987) [PD] Add decode PP event loop for PD disaggregation (sgl-project#14945) ... # Conflicts: # python/sglang/srt/model_executor/piecewise_cuda_graph_runner.py
…14989) Signed-off-by: Ho-Ren (Jack) Chuang <horenchuang@bytedance.com>
…14989) Signed-off-by: Ho-Ren (Jack) Chuang <horenchuang@bytedance.com>
Motivation
#14467 only tested on SM100 and thus didn't test flashmla and fa3. This PR adds the attention backend, flashmla, that can run with FP4 KV cache on SM90.
Experiments
Tested flashmla and fa3 on H20 with DeepSeek-R1-W4AFP8 (MLA) and Qwen3-235B-A22B (MHA).
Checklist