llvm-project

b4df4695 - [CUDA] Implement __CUDA_ARCH_LIST__ macro and refactor architecture helpers (#175260)

[CUDA] Implement __CUDA_ARCH_LIST__ macro and refactor architecture helpers (#175260) Closes #172937 https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/#virtual-architecture-macros > The architecture list macro `__CUDA_ARCH_LIST__` is a list of comma-separated `__CUDA_ARCH__` values for each of the virtual architectures specified in the compiler invocation. The list is sorted in numerically ascending order. Note that unlike NVCC which defines the macro for all C/C++/CUDA compilations done with nvcc, clang defines the macro *only* for CUDA compilations.