[all-commits] [llvm/llvm-project] 2d09be: [AMDGPU] Speedup GCNDownwardRPTracker::advanceBefo...

[Valery via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 2d09bec169277fb5a341249afacff532c7511756
      https://github.com/llvm/llvm-project/commit/2d09bec169277fb5a341249afacff532c7511756
  Author: Valery Pykhtin <valery.pykhtin at gmail.com>
  Date:   2022-12-02 (Fri, 02 Dec 2022)

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

  Log Message:
  -----------
  [AMDGPU] Speedup GCNDownwardRPTracker::advanceBeforeNext

The function makes liveness tests for the entire live register set for every instruction it passes by.
This becomes very slow on high RP regions such as ASAN enabled code.

Instead only uses of last tracked instruction should be tested and this greatly improves compilation time.

This patch revealed few bugs in SIFormMemoryClauses and PreRARematStage::sinkTriviallyRematInsts which should
be fixed first.

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D136267