Fix Windows CUDA build compatibility with newest MSVC #1276
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A recent update to CI build image for Windows has broken the Windows CUDA builds. This is because the latest version of the Visual Studio compiler is version is v1940 [1]. For CUDA versions below 12.4, the check that
nvcc
uses for support is to only allow up to v193x [2]. The intent of that check is to support up to VS2022.There's a change in CMake 3.29.4 for this when using the VS generator [3], but this PR will work around this in a similar way for older CMake versions and when using ninja.
cc: @Titus-von-Koeller
References
[1] https://devblogs.microsoft.com/cppblog/msvc-toolset-minor-version-number-14-40-in-vs-2022-v17-10/
[2] pytorch/vision#8492
[3] https://gitlab.kitware.com/cmake/cmake/-/issues/26003