diff options
| author | Brad King <brad.king@kitware.com> | 2022-03-04 18:51:49 (GMT) |
|---|---|---|
| committer | Brad King <brad.king@kitware.com> | 2022-03-10 14:27:29 (GMT) |
| commit | d1b48bfabd6157309b3056967e6e30cc0ce07983 (patch) | |
| tree | 3ce109b078bf5da4d866f6e2994bbce8246e4df1 /Help/release | |
| parent | 632752d62e8de2730796f509dbb10551351309c2 (diff) | |
| download | CMake-d1b48bfabd6157309b3056967e6e30cc0ce07983.zip CMake-d1b48bfabd6157309b3056967e6e30cc0ce07983.tar.gz CMake-d1b48bfabd6157309b3056967e6e30cc0ce07983.tar.bz2 | |
CUDA: Add support for CUDA_ARCHITECTURES=native
CUDA 11.6 added the `nvcc -arch=native` flag to automatically compile
for the host GPUs' architectures. Add support for specifying this
special `native` value in `CMAKE_CUDA_ARCHITECTURES` and
`CUDA_ARCHITECTURES`. During the compiler ABI detection step,
detect the native architectures so we can pass them explicitly
when using Clang or older versions of nvcc.
Fixes: #22375
Diffstat (limited to 'Help/release')
| -rw-r--r-- | Help/release/dev/cuda-arch-native.rst | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/Help/release/dev/cuda-arch-native.rst b/Help/release/dev/cuda-arch-native.rst new file mode 100644 index 0000000..f44a668 --- /dev/null +++ b/Help/release/dev/cuda-arch-native.rst @@ -0,0 +1,7 @@ +cuda-arch-native +---------------- + +* The :variable:`CMAKE_CUDA_ARCHITECTURES` variable and associated + :prop_tgt:`CUDA_ARCHITECTURES` target property now support the + special ``native`` value to compile for the architectures(s) + of the host's GPU(s). |