NVIDIA / NVIDIA/cutlass

[FEA] Hopper group-gemm of mixed type

Open
#1,614 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

feature request inactive-30d inactive-90d
Dominant language
C++
Stars
10.5k
Forks
2.1k
Avg merge
3d 11h
Merged PRs (30d)
7

Description

Is your feature request related to a problem? Please describe.
For MoE inference, mixed type group-gemm is helpful. But now cutlass seems only support mixed type matmul, and group-gemm of non-mixed type.

Describe the solution you'd like
group-gemm of fp16_int4, fp16_int8, e4m3_int4

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

The request names mixed-type group-gemm support for fp16_int4, fp16_int8, and e4m3_int4, but provides no files, tests, or entry points. Start by locating the existing mixed-type matmul and non-mixed group-gemm implementations, then determine the integration and validation points. Done means the requested Hopper combinations are supported and covered by appropriate tests.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.