Skip to content

[SYCL][HIP] Add gfx9+ hip atomics #8170

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.

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

Merged
merged 21 commits into from
Feb 3, 2023
Merged

Conversation

hdelan
Copy link
Contributor

@hdelan hdelan commented Feb 1, 2023

Adding gfx90a, gfx940 atomic builtins.

@hdelan hdelan requested a review from a team as a code owner February 1, 2023 16:08
@hdelan hdelan requested a review from npmiller February 1, 2023 16:08
@zjin-lcf
Copy link
Contributor

zjin-lcf commented Feb 1, 2023

Just a comment. I ran the sheath-sycl and sheath-hip benchmarks on a MI210 GPU. Thanks.

sycl: Total time for 1000 time steps: 5.1 (s)
hip: Total time for 1000 time steps: 4.97 (s)

@hdelan hdelan temporarily deployed to aws February 1, 2023 17:04 — with GitHub Actions Inactive
@hdelan hdelan temporarily deployed to aws February 1, 2023 18:23 — with GitHub Actions Inactive
@hdelan hdelan temporarily deployed to aws February 1, 2023 19:15 — with GitHub Actions Inactive
@hdelan hdelan force-pushed the newer-hip-atomics branch from b7ecaae to a32a535 Compare February 2, 2023 19:01
@hdelan
Copy link
Contributor Author

hdelan commented Feb 2, 2023

Test suite tests all passing on AMD MI210

@hdelan hdelan temporarily deployed to aws February 2, 2023 19:54 — with GitHub Actions Inactive
@hdelan hdelan temporarily deployed to aws February 2, 2023 21:21 — with GitHub Actions Inactive
@jchlanda jchlanda self-requested a review February 3, 2023 07:07
Copy link
Contributor

@jchlanda jchlanda left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nice one!

@nmnobre
Copy link
Contributor

nmnobre commented Feb 3, 2023

Hi @hdelan,

This is great stuff. 🙌
For my workloads, performance is now on par with hipSYCL and it's actually better than HIP/ROCm!

Any plans to add a toggle like -munsafe-fp-atomics or something similar?

Cheers!

@bader bader merged commit b13561c into intel:sycl Feb 3, 2023
@hdelan
Copy link
Contributor Author

hdelan commented Feb 6, 2023

Hi @nmnobre we will investigate this and add it to the list. Will keep you posted

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants