ci: Fix GPU job not actually enabling OptiX

AcademySoftwareFoundation / OpenShadingLanguage

Advanced shading language for production GI renderers

BSD 3-Clause "New" or "Revised" License

2.05k stars 347 forks source link

ci: Fix GPU job not actually enabling OptiX #1738

Closed brechtvl closed 9 months ago

brechtvl commented 9 months ago

Description

In #1668 USE_OPTIX changed to OSL_USE_OPTIX, but CI was not updated.

And fix CPU SIMD flags being used for CUDA compilation, making the build fail.

Tests

N/A

Checklist:

[x] I have read the contribution guidelines.
[x] I have previously submitted a Contributor License Agreement.
[x] I have updated the documentation, if applicable.
[x] I have ensured that the change is tested somewhere in the testsuite (adding new test cases if necessary).
[x] My code follows the prevailing code style of this project.

AlexMWells commented 9 months ago

 Found CUDA 11.8 
 ...
clang-15: warning: CUDA version is newer than the latest supported version 11.5 [-Wunknown-cuda-version]

So does the clang version need to be paired up with CUDA version? Meaning CUDA 11.5 is newest that could be used when using clang-15? And similar pairings might be true of different clang versions?

brechtvl commented 9 months ago

The GPU job is failing because CPU flags -mavx2 -mf16c are part of the command for building CUDA bitcode, which in turns makes the Imath half.h header try to use CPU intrinsics.

brechtvl commented 9 months ago

That seems to have fixed it. Other CI failures are unrelated.

So does the clang version need to be paired up with CUDA version? Meaning CUDA 11.5 is newest that could be used when using clang-15? And similar pairings might be true of different clang versions?

CUDA 11.8 + LLVM 15 works fine for me despite the warning. But in general, I guess so.