Cold-Air-Networks/mirror-llvm-project
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 7

[Offload] Revert part of #187138. Workaround #191910 (#191925)

Browse source 
Closes #191910

---------

Co-authored-by: Joseph Huber <huberjn@outlook.com>
This commit is contained in:
Ye Luo 2026-04-13 21:39:32 -05:00 committed by GitHub
parent 5753b3f5fd
commit aab5c1075a
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 5 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

5
openmp/device/include/Synchronization.h
View file

@ -62,7 +62,12 @@ V add(Ty *Address, V Val, atomic::OrderingTy Ordering,
template <typename Ty, typename V = utils::remove_addrspace_t<Ty>>
V load(Ty *Address, atomic::OrderingTy Ordering,
MemScopeTy MemScope = MemScopeTy::device) {
#ifdef __NVPTX__
// FIXME: Workaround for older NVIDIA GPUs (#191910).
return __scoped_atomic_fetch_add(Address, V(0), Ordering, MemScope);
#else
return __scoped_atomic_load_n(Address, Ordering, MemScope);
#endif
}
template <typename Ty, typename V = utils::remove_addrspace_t<Ty>>