OpenMathLib / OpenMathLib/OpenBLAS

Modifying GEMM for strided GEMM computation

Open
#2,289 11 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C
Stars
7.6k
Forks
1.7k
Avg merge
1d 3h
Merged PRs (30d)
42

Description

Hi,

We want to implement strided GEMM implementation where the dot product for each output element is computed using only even indexed elements from rows/columns of matrix A/B. Seems like the routine for GEMM is written in assembly? Is there a way to implement this without hacking the assembly? If assembly is the only way can you please help us with this?

Thanks,
Nitish

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

Start by locating the GEMM entry points and assembly routines in OpenBLAS, since the issue does not name a file or test. Determine whether strided access using only even-indexed elements of A and B can be supported without changing assembly. Done means the requested computation is implemented with documented behavior and coverage for the strided case.

Written by the indexing model from the issue text.

Assessment

Tech stack
c
Domain
hpc, performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.