[all-commits] [llvm/llvm-project] 777f91: [AMDGPU] Simplify MergeInfo calculations. NFC.

jayfoad via All-commits all-commits at lists.llvm.org
Wed Apr 29 11:59:41 PDT 2020


  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: 777f91f47e7cd936fb1dfb5c55e87cfd2781e994
      https://github.com/llvm/llvm-project/commit/777f91f47e7cd936fb1dfb5c55e87cfd2781e994
  Author: Jay Foad <jay.foad at amd.com>
  Date:   2020-04-29 (Wed, 29 Apr 2020)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp

  Log Message:
  -----------
  [AMDGPU] Simplify MergeInfo calculations. NFC.

This makes the definition and uses of NewUB more symmetrical, and makes
it clear that ScoreLBs[T] does not change.


  Commit: 120572072eaba4acabcd723317467e20285c4479
      https://github.com/llvm/llvm-project/commit/120572072eaba4acabcd723317467e20285c4479
  Author: Jay Foad <jay.foad at amd.com>
  Date:   2020-04-29 (Wed, 29 Apr 2020)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp

  Log Message:
  -----------
  [AMDGPU] Initialize gpr upper bounds to -1. NFC.

These upper bounds are inclusive, so -1 (rather than 0) is the natural
way to express an empty range.


  Commit: 31c09d03a1fb58f96417e4bfed4d1e5a82124606
      https://github.com/llvm/llvm-project/commit/31c09d03a1fb58f96417e4bfed4d1e5a82124606
  Author: Jay Foad <jay.foad at amd.com>
  Date:   2020-04-29 (Wed, 29 Apr 2020)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp

  Log Message:
  -----------
  [AMDGPU] Remove WaitcntBrackets::MixedPendingEvents[]. NFC.

It's trivial to derive this information from other state.


Compare: https://github.com/llvm/llvm-project/compare/4c53f4202a45...31c09d03a1fb


More information about the All-commits mailing list