-
-
Notifications
You must be signed in to change notification settings - Fork 20
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
Test CUDA compilation on Windows #115
Conversation
@conda-forge-admin, please rerender |
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( I do have some suggestions for making it better though... For recipe:
|
…nda-forge-pinning 2024.05.08.04.59.39
@conda-forge-admin, please rerender |
…nda-forge-pinning 2024.05.08.04.59.39
CUDA 12 error:
Usual CUDA 12 migration problem. CUDA 11.8 error:
|
Same error of #63 (comment), however now CUDA 11.8 should still be supported. Anyhow, it make sense to first concentrate on CUDA 12.0 . |
Based on:
I guess CUDA_HOME can be simply |
@conda-forge-admin, please rerender |
Now the builds (also the non-CUDA one) fails with:
this smells like an environment regression, for example in CI images or similar. |
…nda-forge-pinning 2024.05.16.11.05.29
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( I do have some suggestions for making it better though... For recipe:
|
Also CUDA 12 fails with:
|
Probably there is some argument that should not be there, that is then interpreter by the compiler as an input file, generating the error (see https://forums.developer.nvidia.com/t/solved-nvcc-fatal-a-single-input-file-is-required-for-a-non-link-phase-when-an-outputfile-is-s/37807 for a similar error). I can easily check this locally to understand which command is wrong. |
By the way, this was fixed by conda-forge/vc-feedstock#76, thanks @mgovers! |
The problem in:
Are the |
As it does not seem to me that |
This fix worked fine. Now the CUDA 12 builds fail (at test level) as it run out of disk space, while CUDA 11 runs out of time. We can switch to |
Apparently free_disk_space is only available on Linux. Let's just switch to all-major for now and see how it goes. |
I remember that there was another trick on Windows: f3662ef . Let's try to use that an still use |
@conda-forge-admin please rerender |
…nda-forge-pinning 2024.05.21.18.33.22
Not sure why, but now all the jobs (even the non-cuda one) are failing with MemoryError while linking
|
Maybe we can skip making the test executable…. |
Yes, this is exactly what is happening now. I would still like to run tests on non-cuda builds, but that is not a blocker. |
ah very cool! |
Ok, we have a solution working, let me check if I can improve this, but if I fail I will go back to this and cleanup the PR (eventually rebasing on the 1.18.0 one). |
Hi! This is the friendly automated conda-forge-linting service. I was trying to look for recipes to lint for you, but it appears we have a merge conflict. Please ping the 'conda-forge/core' team (using the @ notation in a comment) if you believe this is a bug. |
Follow up and full matrix test in #121 . |
Checklist
0
(if the version changed)conda-smithy
(Use the phrase@conda-forge-admin, please rerender
in a comment in this PR for automated rerendering)