[Openmp-commits] [PATCH] D104666: [Libomptarget] Improve device runtime implementation for globalized variables.

Johannes Doerfert via Phabricator via Openmp-commits openmp-commits at lists.llvm.org
Mon Jun 21 13:41:46 PDT 2021


jdoerfert added inline comments.


================
Comment at: openmp/libomptarget/deviceRTLs/common/src/data_sharing.cu:42
+
+[[clang::loader_uninitialized]] ThreadStackTy<MinBytes * 8, 1> MainSharedStack;
+#pragma omp allocate(MainSharedStack) allocator(omp_pteam_mem_alloc)
----------------
tianshilei1992 wrote:
> IIRC, in the new `deviceRTLs`, we only have one stack where the first chunk, which is bigger than the rest, is for the main thread in non-SPMD mode. Why do we want to have two here?
Was easier to write it like this from scratch, either way works, no real difference, this might be actually nicer.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D104666/new/

https://reviews.llvm.org/D104666



More information about the Openmp-commits mailing list