[all-commits] [llvm/llvm-project] 13877d: [AMDGPU] Fix shortfalls in WQM marking

[Carl Ritson via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 13877db2fa859c9753bd87a47ef00cf262516ae7
      https://github.com/llvm/llvm-project/commit/13877db2fa859c9753bd87a47ef00cf262516ae7
  Author: Carl Ritson <carl.ritson at amd.com>
  Date:   2021-03-15 (Mon, 15 Mar 2021)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIWholeQuadMode.cpp
    M llvm/test/CodeGen/AMDGPU/wqm.ll
    M llvm/test/CodeGen/AMDGPU/wqm.mir

  Log Message:
  -----------
  [AMDGPU] Fix shortfalls in WQM marking

When tracking defined lanes through phi nodes in the live range
graph each branch of the phi must be handled independently.
Also rewrite the marking algorithm to reduce unnecessary
operations.

Previously a shared set of defined lanes was used which caused
marking to stop prematurely. This was observable in existing lit
tests, but test patterns did not cover this detail.

Reviewed By: piotr

Differential Revision: https://reviews.llvm.org/D98614