Hi, According to https://stackoverflow.com/questions/20186848/can-i-compile-a-cuda-program-without-having-a-cuda-device and quick attempt to install nvidia-cuda-toolkit-12.3.0.ebuild with removed x11-drivers/nvidia-drivers dependency, cuda compiler works fine without CUDA driver and NVIDIA GPU. As a developer, I just want to check that my code compiles code on my machine and then occasionally test compiled binaries somewhere else (Google Colab, Saturn labs, etc). Additionally, tools like hipify-clang require only CUDA headers, even if the target platform is AMD. Could you unbind dev-util/nvidia-cuda-toolkit from x11-drivers/nvidia-drivers? Thanks!
Unbinding dev-util/nvidia-cuda-toolkit from x11-drivers/nvidia-drivers may be interesting for using cuda inside WSL2.
I support removing the dependency as well. It might also be useful on Gentoo prefix environments.