[PATCH] D75788: [WIP][OpenMP] Reuse CUDA wrappers in `nvptx` target regions.

[Johannes Doerfert via Phabricator via cfe-commits]

jdoerfert created this revision.
jdoerfert added reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim, aaron.ballman.
Herald added subscribers: guansong, bollu, mgorny.
Herald added a project: clang.

This is a WIP patch to show one way of implementing math support (and
support for `new`, `algorithm`, ...) in OpenMP target regions by piggy
backing on the CUDA implementation. All CUDA "code" should be in
`omp begin/end declare variant match(device={arch(nvptx)})` at the end
of the day while the "host" code is not.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D75788

Files:
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Headers/CMakeLists.txt
  clang/lib/Headers/__clang_cuda_cmath.h
  clang/lib/Headers/__clang_cuda_device_functions.h
  clang/lib/Headers/__clang_cuda_libdevice_declares.h
  clang/lib/Headers/__clang_cuda_math_forward_declares.h
  clang/lib/Headers/__clang_cuda_runtime_wrapper.h
  clang/lib/Headers/cuda_wrappers/new
  clang/lib/Headers/openmp_wrappers/__clang_openmp_math.h
  clang/lib/Headers/openmp_wrappers/__clang_openmp_math_declares.h
  clang/lib/Headers/openmp_wrappers/cmath
  clang/lib/Headers/openmp_wrappers/math.h
  clang/test/OpenMP/target_nvptx_math_fp_macro.cpp
  clang/test/OpenMP/target_nvptx_math_sin.c

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D75788.248864.patch
Type: text/x-patch
Size: 23454 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20200307/14fb5586/attachment-0001.bin>