[all-commits] [llvm/llvm-project] 8f6c47: [AMDGPU] Speedup GCNDownwardRPTracker::advanceBefo...

[Valery via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 8f6c47b7a42cedffc1f2ec7d391385a34d1d5569
      https://github.com/llvm/llvm-project/commit/8f6c47b7a42cedffc1f2ec7d391385a34d1d5569
  Author: Valery Pykhtin <valery.pykhtin at gmail.com>
  Date:   2023-03-09 (Thu, 09 Mar 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/GCNRegPressure.cpp
    A llvm/test/CodeGen/AMDGPU/scheduler-rp-calc-one-successor-two-predecessors-bug.ll

  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