-
-
Notifications
You must be signed in to change notification settings - Fork 5.4k
TST: linalg: race condition in BLAS ger*
#23084
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
Conversation
Thanks, glad to see that fixed now. We should only merge this after NumPy 2.3.0 is out with the fix. |
NumPy 2.3.0 final is out and this should be now mergeable (if CI agrees) |
ger*
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
CI is happy, but that's expected since parallel testing is currently disabled. Some local testing shows that there's still a problem, this crashes regularly:
pixi r test-nogil -t scipy.linalg.tests.test_blas -- --parallel-threads=5 -k test_ger
Can't reproduce on my box on 32 threads. Could you |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I can no longer reproduce the issue after thoroughly cleaning build env and caches - not sure what happened there. In it goes, thanks @crusaderky!
*ger*
routines inlinalg.blas
are not thread-safe #21936Tested many times on a 32-CPUs host with
--parallel-threads=32
.