HomePhabricator

[AMDGPU] Fix shortfalls in WQM marking

Authored by critson on Mar 15 2021, 5:32 AM.

Description

[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

Details

Committed
critsonMar 15 2021, 5:44 AM
Reviewer
piotr
Differential Revision
D98614: [AMDGPU] Fix shortfalls in WQM marking
Parents
rG07232f4507e8: [X86][SSE] canonicalizeShuffleWithBinOps - add X86ISD::PSHUFB handling.
Branches
Unknown
Tags
Unknown