New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Small GEMM improvements for AArch64 with SVE #4887

Merged

martin-frbg merged 1 commit into OpenMathLib:develop from goplanid:develop

Sep 16, 2024

Contributor

goplanid commented Sep 2, 2024 •

edited

Loading

This pull request improves dgemm small matrix kernel of TN case for for AArch64 by unrolling the columns further.

Results are demonstrated with below 2 graphs:

There is improvement seen in the graph stability for square matrix shapes.
For irregular matrix shape (primarily issue seen when M is changing with N and K constant), there is 25% improvement.

Note: These results were generated on an AWS Graviton 3 machine with 32 cores.


          Improve TN case with further unrolling

4894c54

Contributor Author

goplanid commented Sep 3, 2024

cc: @Mousius . This is in continuation of PR #4806

martin-frbg added this to the 0.3.29 milestone

martin-frbg merged commit fca86e3 into OpenMathLib:develop

76 of 78 checks passed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet