[clang] [llvm] [Clang][AMDGPU] Add nontemporal scoped atomic load/store and merge relaxed atomics (PR #229207)

Matt Arsenault via cfe-commits cfe-commits at lists.llvm.org
Tue Oct 6 01:21:31 PDT 2026


arsenm wrote:

Didn't we add the cooperative atomics builtins for the rccl case? Why aren't they using that? 

https://github.com/llvm/llvm-project/pull/229207


More information about the cfe-commits mailing list