Skip to content

[CUDA] update default CUDA sm ver to 75 #20824

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
Open
bratpiorka wants to merge 5 commits into
base: sycl
Choose a base branch
from
Open

[CUDA] update default CUDA sm ver to 75 #20824

bratpiorka wants to merge 5 commits into from

Conversation

@bratpiorka
Copy link
Contributor

@bratpiorka bratpiorka commented Dec 4, 2025 •
edited
Loading

This pull request updates the default NVIDIA GPU architecture used for CUDA from SM_50 to SM_75, as the newest CUDA 13 no longer supports SM_50.

It also updates several tests:

  • sets CUDA 10.0 as the default toolkit version used in driver-detection tests
  • bumps the SM version to 75 and the PTX version to 63 for tests that look for specific output patterns
  • adds support for the 3-operand atomic intrinsic
  • adds support for the tanh.approx.f16/f16x2 intrinsic

@bratpiorka bratpiorka changed the title [WIP] update default CUDA sm ver to 75 [CUDA] update default CUDA sm ver to 75 Dec 4, 2025
steffenlarsen
steffenlarsen reviewed Dec 4, 2025
View reviewed changes
Copy link
Contributor

@steffenlarsen steffenlarsen left a comment

Choose a reason for hiding this comment

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

Overall, I think we should get most of these changes into upstream to avoid the headaches of maintaining customizations in intel/llvm, especially for changes that are not directly SYCL-related.

clang/include/clang/Basic/OffloadArch.h
LAST,

CudaDefault = OffloadArch::SM_52,
CudaDefault = OffloadArch::SM_75,
Copy link
Contributor

@steffenlarsen steffenlarsen Dec 4, 2025

Choose a reason for hiding this comment

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

A bunch of this are customizations to upstream LLVM. Even if we go for it downstream, it would be good to start the process of upstreaming these customizations.

Copy link
Contributor

@sarnex sarnex Dec 4, 2025

Choose a reason for hiding this comment

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

+1

Copy link
Contributor Author

@bratpiorka bratpiorka Dec 5, 2025

Choose a reason for hiding this comment

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

the upstream version is here: llvm/llvm-project#170679

mdtoguchi
mdtoguchi reviewed Dec 4, 2025
View reviewed changes
bader
bader previously requested changes Dec 4, 2025
View reviewed changes
Copy link
Contributor

@bader bader left a comment

Choose a reason for hiding this comment

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

@bratpiorka, please remove any changes from mlir/ project files. We don't support this project and these modifications impact pulldown from the upstream.

@bratpiorka
Copy link
Contributor Author

bratpiorka commented Dec 5, 2025

@bratpiorka, please remove any changes from mlir/ project files. We don't support this project and these modifications impact pulldown from the upstream.

@bader thanks, removed

@bader bader dismissed their stale review December 5, 2025 17:52

All changes from the files I own has been reverted.

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

Reviewers

@sarnex sarnex sarnex left review comments

@mdtoguchi mdtoguchi mdtoguchi left review comments

@steffenlarsen steffenlarsen steffenlarsen left review comments

@bader bader bader left review comments

@cperkinsintel cperkinsintel Awaiting requested review from cperkinsintel cperkinsintel is a code owner

@aelovikov-intel aelovikov-intel Awaiting requested review from aelovikov-intel aelovikov-intel is a code owner

At least 1 approving review is required to merge this pull request.

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@bratpiorka @bader @sarnex @mdtoguchi @steffenlarsen