Skip to content

CUDA: don't device link if target has no CUDA usage

When CUDA is enabled, and a pure non-CUDA target has CMAKE_CUDA_SEPARABLE_COMPILATION enabled, don't actually perform the device linking step, as it will fail. A target that has CMAKE_CUDA_SEPARABLE_COMPILATION enabled must also have CUDA usage (either itself, or something it links to).

Fixes: #20182 (closed)
Topic-rename: cuda-device-link-only-cuda
Backport: release:HEAD^2

Edited by Brad King

Merge request reports