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

__init_driver__ not called in 3.5 #1280

Closed
mwarusz opened this issue Dec 16, 2021 · 1 comment · Fixed by #1287
Closed

__init_driver__ not called in 3.5 #1280

mwarusz opened this issue Dec 16, 2021 · 1 comment · Fixed by #1287
Labels
bug Something isn't working

Comments

@mwarusz
Copy link
Contributor

mwarusz commented Dec 16, 2021

I noticed a substantial slowdown (30%) of one of my codes after upgrading CUDA.jl to 3.5.
I bisected it to 0bc1191.
It seems that after this commit the __init_driver__ function is no longer called and the
-nvptx-fma-level=1 LLVM option is not set, inhibiting fma generation.

@lcw

@mwarusz mwarusz added the bug Something isn't working label Dec 16, 2021
@maleadt
Copy link
Member

maleadt commented Dec 20, 2021

Thanks for the bisect! Re-introduced the call in #1287.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants