vllm-project/vllm

[Feature]: Integrate RMS+fp4 fused kernel from FlashInfer

Open

#32612 opened on Jan 19, 2026

View on GitHub
 (14 comments) (0 reactions) (1 assignee)Python (80,034 stars) (16,816 forks)batch import
feature requestgood first issuehelp wantedstaletorch.compile

Description

🚀 The feature, motivation and pitch

Kernel: https://github.com/flashinfer-ai/flashinfer/blob/main/flashinfer/norm.py#L406-L409

This should be integrated in the existing rms + quant fusion pass using a similar approach as the silu+fp4 fusion in the act quant fusion pass.

Alternatives

No response

Additional context

No response

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Contributor guide

Tech stack
pythoncpp
Domain
backendperformance
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
blocked
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
PythonC++/CUDAFused kernel conceptsQuantization
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
15
Research direction
Study the existing RMS+quant fusion pass in the vllm codebase and the analogous silu+fp4 fusion in the act quant fusion pass. Review the FlashInfer RMS+fp4 kernel implementation at flashinfer/norm.py#L406 L409. Identify integration points and potential modifications. Engage with the assignee in the issue comments to clarify scope and avoid redundant work.