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

Non-transpose gemv should use TeamPolicy on GPUs #926

Closed
brian-kelley opened this issue Apr 1, 2021 · 1 comment
Closed

Non-transpose gemv should use TeamPolicy on GPUs #926

brian-kelley opened this issue Apr 1, 2021 · 1 comment
Assignees
@brian-kelley
Labels
enhancement InDevelop

Comments

@brian-kelley
Copy link
Contributor

brian-kelley commented Apr 1, 2021 •
edited
Loading

Our implementation of dense mat-vec (gemv) uses a range policy (single level of parallelism) on all execution spaces. It should be using a TeamPolicy since it's trivial to parallelize the sums within each row.
@brian-kelley brian-kelley self-assigned this Apr 1, 2021
@brian-kelley
Copy link
Contributor Author

This should be benchmarked for local matrices sized about 1e4 x 1e5, since this is getting used on large row-partitioned square matrices.

@brian-kelley brian-kelley mentioned this issue Apr 20, 2021
Merged
brian-kelley added a commit that referenced this issue Apr 27, 2021
@brian-kelley
Merge pull request #939 from brian-kelley/TeamGEMV
73fbbc8 
Add fast two-level mode N GEMV (#926)
@kokkos-devops-admin kokkos-devops-admin mentioned this issue Feb 20, 2024
Merged
@kokkos-devops-admin kokkos-devops-admin mentioned this issue May 24, 2024
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@brian-kelley brian-kelley

Labels
enhancement InDevelop
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@brian-kelley @ndellingwood