[Openmp-commits] [PATCH] D135444: [OpenMP] Utilize the "non-uniform-workgroup" to simplify DeviceRTL

Ethan Stewart via Phabricator via Openmp-commits openmp-commits at lists.llvm.org
Wed Nov 2 07:45:07 PDT 2022


estewart08 added inline comments.


================
Comment at: openmp/libomptarget/DeviceRTL/src/Mapping.cpp:82
 
-uint32_t getNumberOfBlocks() {
-  return getGridDim(__builtin_amdgcn_grid_size_x(),
-                    __builtin_amdgcn_workgroup_size_x());
-}
+uint32_t getNumberOfBlocks() { return __builtin_amdgcn_grid_size_x(); }
 
----------------
jdoerfert wrote:
> This should be `__builtin_amdgcn_grid_size_x()/__builtin_amdgcn_workgroup_size_x()`. 
That fixed the check-openmp failures on my local build.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D135444/new/

https://reviews.llvm.org/D135444



More information about the Openmp-commits mailing list