[all-commits] [llvm/llvm-project] 13b63b: [AMDGPU] prefer non-mfma in post-RA schedule

[Stanislav Mekhanoshin via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: 13b63be472233762024ef196dd88612369a51807
      https://github.com/llvm/llvm-project/commit/13b63be472233762024ef196dd88612369a51807
  Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
  Date:   2020-07-29 (Wed, 29 Jul 2020)

  Changed paths:
    M llvm/lib/Target/AMDGPU/GCNHazardRecognizer.cpp
    M llvm/lib/Target/AMDGPU/GCNHazardRecognizer.h
    A llvm/test/CodeGen/AMDGPU/sched-prefer-non-mfma.mir

  Log Message:
  -----------
  [AMDGPU] prefer non-mfma in post-RA schedule

MFMA instructions shall not be scheduled back to back
to avoid MAI SIMD stall. Tell post-RA schedule we would
prefer some other instruction instead.

Differential Revision: https://reviews.llvm.org/D84883