[Openmp-commits] [PATCH] D138389: [OpenMP][libomptarget] Add AMDGPU NextGen plugin with asynchronous behavior

[Kevin Sala Penadés via Phabricator via Openmp-commits]

kevinsala updated this revision to Diff 482161.
kevinsala added a comment.

Rebasing onto `main` to include several pre-commits. Fixing several reviewer's issues. Also making streams dynamically sized (`std::deque` to avoid invalidating elements on insertions). In this way, we can re-use signals to reduce signal consumption through a signal resource manager.

There are some issues to fix:

- There is still one error when running the tests. The stream hangs with the recent changes regarding dynamically sized streams. Currently under investigation.
- Missing documentation on openmp/docs. We can open a separate patch for that.
- Missing std::atomic_thread_fence inside the stream implementation. May be related to the issue mentioned above.
- Some reviewer's comments not fixed yet. Currently working on it.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D138389/new/

https://reviews.llvm.org/D138389

Files:
  openmp/libomptarget/plugins-nextgen/CMakeLists.txt
  openmp/libomptarget/plugins-nextgen/amdgpu/CMakeLists.txt
  openmp/libomptarget/plugins-nextgen/amdgpu/src/rtl.cpp
  openmp/libomptarget/plugins/amdgpu/dynamic_hsa/hsa.h

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D138389.482161.patch
Type: text/x-patch
Size: 103402 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/openmp-commits/attachments/20221212/4ad4678c/attachment-0001.bin>