Improves the handling of CUDA layouts where we have multiple include
and library directories listed in the output of `nvcc -v`. This
updates both when the CUDA language is enabled or not.
Fixes: #24915
Update the Clang version checks added by
* commit eacacc70fc (Clang: use -std=c++23 for Clang 17.0+,
2023-05-08, v3.27.0-rc1~115^2)
* commit 0183956d30 (Clang: C++26 support,
2023-05-16, v3.27.0-rc1~85^2)
to account for Android NDK r26's distribution of a Clang based on a
development version of LLVM/Clang 17.0 that pre-dated addition of
these flags.
Follow the pattern from commit 12e6796b62 (Android: Do not pass
non-existent Clang -std flags, 2018-04-02, v3.12.0-rc1~284^2),
which fixed the same problem for `-std=c++17`.
Fixes: #25281
ROCm may not install the `hip-lang` cmake package on NVIDIA platforms.
Skip verifying its presence. We do not use it for NVIDIA platforms anyway.
Fixes: #25271
4089954595 gitlab-ci: update to GCC CI container with merged P1689R5
c2564f5916 ci: drop extra scandep source settings for GCC
ee52a02d56 ci: update GCC build to use master after P1689R5 is merged
4f0410959e GCC: add rules for scanning C++ module dependencies
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !8827
I was noticing that FindOpenMP was failing to find my openmp installation
on macos when building with a custom-compiled clang. Upon inspection it
appears that FindOpenMP is looking for `AppleClang` as a compiler
identification and the `Clang` case is only applicable for Win32.
Try to fix that by treating custom `Clang` on macos the same as
`AppleClang`.
This makes detection of Doxygen more robust in case of broken Doxygen
executables. Doxygen is called to determine its version number. If this
call returns a non-zero result, we have an indication that the Doxygen
executable is broken. We report Doxygen as not found to prevent issues
with checks for other Doxygen component, that might lead to
configuration aborts.
Fixes: #18708
6546aa2a2a ci: Add HIP job using CUDA on NVIDIA GPUs
18158bf81c HIP: Add support for NVIDIA GPUs
127b6fa06b HIP: Add CMAKE_HIP_PLATFORM variable to specify GPU platform
90e23f40ee Tests/HIP/WithDefs: Clean up test case
9ebdf3281f Tests/HIP/ArchitectureOff: Cover HIP_ARCHITECTURES initialization
cfec29196e ci: Add CUDA 11.8 to HIP 5.5 image
26470eb987 ci: Put HIP GPU platform in CMake_TEST_HIP
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !8817
b32ea7cff0 CUDA: Factor out helper to generate CUDA architecture flags
b3a1f17567 CUDA: Factor out helper to validate CMAKE_CUDA_ARCHITECTURES
8617c28221 CUDA: Factor out helper for detecting native CUDA architectures
0db0fe7958 CUDA: Factor out helper to compute all CUDA architectures
c16f1e2b93 CUDA: Factor out helper to find CUDA Toolkit for compiler
a6841a967b CUDA: Factor out helper to filter implicit link libraries
deff0e638d CUDA: Factor out helper to parse NVCC implicit compiler and linker flags
e1b2a5062f CUDA: Factor out some NVCC compiler information
...
Acked-by: Kitware Robot <kwrobot@kitware.com>
Reviewed-by: Raul Tambre <raul@tambre.ee>
Merge-request: !8816
Improve variable names and comments.
Remove a comment missed by commit c267ed205a (CUDA: Defer architecture
testing to the compiler testing step, 2022-04-25, v3.24.0-rc1~222^2)
about architecture verification.
Robert Maynard