[Openmp-commits] [PATCH] D125698: [Libomptarget] Don't build the device runtime without a new Clang

Mon May 16 12:52:10 PDT 2022

ye-luo added inline comments.

================
Comment at: openmp/libomptarget/DeviceRTL/CMakeLists.txt:253
 # Set the flags to build the device runtime from clang.
 set(clang_lib_flags -fopenmp -fopenmp-cuda-mode -foffload-lto -fvisibility=hidden -Xopenmp-target=nvptx64-nvidia-cuda --cuda-feature=+ptx61 -mllvm -openmp-opt-disable -nocudalib -nogpulib -nostdinc -DSHARED_SCRATCHPAD_SIZE=512)
 foreach(arch ${nvptx_sm_list})
----------------
54e02179b33f6bfd1c7bd836e790effd5ffb715c changed this line and `-Xopenmp-target=nvptx64-nvidia-cuda --cuda-feature=+ptx61` doesn't seem making sense for amdgpu.

================
Comment at: openmp/libomptarget/DeviceRTL/CMakeLists.txt:262
 # Build the static library version of the device runtime.
 add_library(omptarget.devicertl STATIC)
 target_compile_features(omptarget.devicertl PRIVATE cxx_std_17)
----------------
This target doesn't get added like other custom target is likely the source of all the problems.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125698/new/

https://reviews.llvm.org/D125698