Skip to content

alibaba/redfuser

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13,374 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RedFuser

License ASPLOS 2026

This repository contains the artifact for our ASPLOS 2026 paper: "RedFuser: An Automatic Operator Fusion Framework for Cascaded Reductions on AI Accelerators".

Overview

RedFuser is a novel framework for optimizing cascaded reductions in deep learning compilers. Built on top of Apache TVM, RedFuser introduces a series of compiler transformation passes that enable efficient fusion of reduction operations with other computations, particularly targeting modern GPU architectures.

Update

  • [2026-01]: RedFuser is now avaliable with flash-attention example.
  • [2025-11]: 🎉RedFuser is accepted by ASPLOS 2026!

Roadmap

  • flash-attention
  • flash-decoding
  • moe-routing
  • fp8 quant+gemm

Getting Started

Please follow https://tvm.apache.org/docs/install/index.html to install.

Example

For flash-attention example, see python/tvm/redfuser/example/flash_attention.py.

Structure

redfuser/
├── python/tvm/redfuser/       # RedFuser Python implementation
│   ├── transform/             # Core transformation passes
│   └── example/               # Example workloads
│ ...

License

RedFuser is licensed under the Apache License 2.0.

Acknowledgments

This project builds upon Apache TVM. We thank the TVM community for their excellent infrastructure and support.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors