Skip to content

Partially revert c76140fa277895b3886cf1780a36fe504a1aac3a: revert amdsmi-related cmake changes only#4923

Open
araravik-psd wants to merge 1 commit intodevelopfrom
users/arravikum/partial-revert-4802
Open

Partially revert c76140fa277895b3886cf1780a36fe504a1aac3a: revert amdsmi-related cmake changes only#4923
araravik-psd wants to merge 1 commit intodevelopfrom
users/arravikum/partial-revert-4802

Conversation

@araravik-psd
Copy link
Copy Markdown
Contributor

@araravik-psd araravik-psd commented Apr 10, 2026

This PR is intended to partially land back reverted changes in c76140f other than the amdsmi cmake changes which were landed as part of the PR.

The Primary revert #4802 was originally applied to get Pytorch builds back to green as explained in #4802 description.

Adding back changes for thread-safety and wrapper usage as requested in the issue comment #4802 (comment)

These changes were tested with a submodule bump in TheRock:

Passing Pytorch builds:
https://github.com/ROCm/TheRock/actions/runs/24211884393
https://github.com/ROCm/TheRock/actions/runs/24208190701

@araravik-psd araravik-psd requested a review from a team as a code owner April 10, 2026 20:12
@ScottTodd ScottTodd removed their request for review April 10, 2026 20:12

// Enable use of amd_smi_lib instead of internal ARSMI code by default; set RCCL_USE_AMD_SMI_LIB=1 to enable amd_smi_lib and use the AMD SMI path
RCCL_PARAM(UseAmdSmiLib, "USE_AMD_SMI_LIB", AMDSMI_DEFAULT_ENABLED);
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

This needs to be set to disabled by default

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

@mustafabar Please take a look

@math-ci
Copy link
Copy Markdown

math-ci bot commented Apr 10, 2026

regression-detection smoke (multi-node) on commit 85819f4

Artifacts - No regressions preview generated.

@math-ci
Copy link
Copy Markdown

math-ci bot commented Apr 10, 2026

regression-detection smoke (single-node) on commit 85819f4

Artifacts - No regressions preview generated.

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

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants