[all-commits] [llvm/llvm-project] f238a9: [OpenMP][libomptarget][AMDGPU] Enable active HSA w...

[Greg Rodgers via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: f238a98e844752b955dcf3d7b95b9c76c75a0017
      https://github.com/llvm/llvm-project/commit/f238a98e844752b955dcf3d7b95b9c76c75a0017
  Author: gregrodgers <Gregory.Rodgers at amd.com>
  Date:   2023-05-04 (Thu, 04 May 2023)

  Changed paths:
    M openmp/docs/design/Runtimes.rst
    M openmp/libomptarget/plugins-nextgen/amdgpu/src/rtl.cpp

  Log Message:
  -----------
  [OpenMP][libomptarget][AMDGPU] Enable active HSA wait state

Adds HSA timeout hint of 2 seconds to the AMDGPU nextgen-plugin to improve
performance of small kernels.
The HSA runtime may stay in HSA_WAIT_STATE_ACTIVE for up to the timeout
value before switching to HSA_WAIT_STATE_BLOCKED. This can improve
latency from which small kernels can benefit.
The value was determined via experimentation w/ different benchmarks.

The timeout value can be overriden using the environment variable
LIBOMPTARGET_AMDGPU_STREAM_BUSYWAIT with a value in microseconds.

Original author: Greg Rodgers <Gregory.Rodgers at amd.com>
Contributions from: JP Lehr <JanPatrick.Lehr at amd.com>

Differential Revision: https://reviews.llvm.org/D148808