Closed fel-cab closed 2 months ago
@llvm/issue-subscribers-openmp
Author: Felipe Cabarcas (fel-cab)
@llvm/issue-subscribers-offload
Author: Felipe Cabarcas (fel-cab)
@llvm/issue-subscribers-bug
Author: Felipe Cabarcas (fel-cab)
This is a backend issue.
@llvm/issue-subscribers-backend-amdgpu
Author: Felipe Cabarcas (fel-cab)
This code cannot reliably work. You are trying to synchronize within a warp on an AMDGPU, that does not work. OpenMP 6.0 makes this explicit. If you want to test atomics, split the reader and writer into different warps at least.
On a MI250x
The following program doesn't finish when compiled as:
clang -fopenmp -O3 --offload-arch=gfx90a
But it does finish when compiled as:
clang -fopenmp --offload-arch=gfx90a