[all-commits] [llvm/llvm-project] 7c9c6c: clang/AMDGPU: Set amdgpu-max-num-workgroups to dis...

[Matt Arsenault via All-commits]

  Branch: refs/heads/users/arsenm/clang-amdgpu-set-default-gridsize-1d-dispatch
  Home:   https://github.com/llvm/llvm-project
  Commit: 7c9c6c6897c74df7427c20b0e1305efaa2028c8c
      https://github.com/llvm/llvm-project/commit/7c9c6c6897c74df7427c20b0e1305efaa2028c8c
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2024-12-06 (Fri, 06 Dec 2024)

  Changed paths:
    M clang/include/clang/Basic/LangOptions.h
    M clang/lib/CodeGen/Targets/AMDGPU.cpp
    M clang/test/CodeGenHIP/default-attributes.hip
    M clang/test/OpenMP/amdgcn-attributes.cpp
    M clang/test/OpenMP/amdgcn_target_global_constructor.cpp

  Log Message:
  -----------
  clang/AMDGPU: Set amdgpu-max-num-workgroups to disable Y/Z by default

Only OpenCL supports 2d and 3d dispatches, the other languages Y and Z
dimensions are always 1. Some of the generated OpenMP functions don't
seem to get the correct attributes. The kernels do, but the callable
__omp_offloading functions are missing them for some reason.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications