[NVPTX, CUDA] Add .and.popc variant of the b1 MMA instruction.
That should allow clang to compile mma.h from CUDA-11.3. Differential Revision: https://reviews.llvm.org/D105384
Loading
Please sign in to comment
That should allow clang to compile mma.h from CUDA-11.3. Differential Revision: https://reviews.llvm.org/D105384