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.

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

Widen gpu allocations buffer #2878

Merged
merged 1 commit into from
Apr 6, 2024
Merged

Conversation

charleskawczynski
Copy link
Member

This PR widens the gpu allocations buffer, so that it fails less frequently.

@Sbozzolo
Copy link
Member

Sbozzolo commented Apr 5, 2024

The job still failed. :(

Also, having a buffer that is 5 times larger than the target one seems to defy the purpose of this job.

@charleskawczynski
Copy link
Member Author

Also, having a buffer that is 5 times larger than the target one seems to defy the purpose of this job.

Agreed, but the flakiness is a drag on CI. I think I'd prefer to remove the lower limit, and just add an info statement, like we used to have.

@Sbozzolo
Copy link
Member

Sbozzolo commented Apr 5, 2024

If this test is barely indicative, let us remove it or have it soft fail. We can revisit it with julia 1.11.

@charleskawczynski
Copy link
Member Author

If this test is barely indicative, let us remove it or have it soft fail. We can revisit it with julia 1.11.

The problem with soft fail is that it can then fail for other reasons, and CI will still pass. I'd prefer we just relax the limits, so that CI is no longer flakey. I'll bump it 10x, and remove the lower limit, for the gpu job.

@charleskawczynski charleskawczynski added this pull request to the merge queue Apr 6, 2024
Merged via the queue into main with commit d57b5ce Apr 6, 2024
11 checks passed
@charleskawczynski charleskawczynski deleted the ck/gpu_allocs_buffer branch April 6, 2024 02:19
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.

2 participants