Skip to content
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

Enable context caching for cuBLAS backend to improve performance #111

Closed
mmeterel opened this issue Jul 13, 2021 · 0 comments · Fixed by #169
Closed

Enable context caching for cuBLAS backend to improve performance #111

mmeterel opened this issue Jul 13, 2021 · 0 comments · Fixed by #169

Comments

@mmeterel
Copy link
Contributor

Summary

As noted in #106, performance with cuBLAS backend is low and can be improved if CUDA context is cached.
Opening this issue as a tracker.

Version

Appears in latest.

Environment

cuBLAS backend

Steps to reproduce

See #106

Observed behavior

Running cuBLAS backend through oneMKL is slower than running cuBLAS directly.

Expected behavior

Running cuBLAS backend through oneMKL should match or perform very close to pure cuBLAS.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

[BLAS] Fix cublas perf aelizaro/oneMKL
1 participant
@mmeterel