[all-commits] [llvm/llvm-project] c80d8a: [AMDGPU] MachineLICM cannot hoist VALU

[Stanislav Mekhanoshin via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: c80d8a8ceabb5c42e0143d9507394068075538f9
      https://github.com/llvm/llvm-project/commit/c80d8a8ceabb5c42e0143d9507394068075538f9
  Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
  Date:   2021-10-20 (Wed, 20 Oct 2021)

  Changed paths:
    M llvm/lib/CodeGen/MachineLoopInfo.cpp
    M llvm/test/CodeGen/AMDGPU/dag-divergence-atomic.ll
    M llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll
    M llvm/test/CodeGen/AMDGPU/global-atomics-fp.ll
    M llvm/test/CodeGen/AMDGPU/indirect-call.ll
    M llvm/test/CodeGen/AMDGPU/infinite-loop.ll
    M llvm/test/CodeGen/AMDGPU/mul24-pass-ordering.ll
    M llvm/test/CodeGen/AMDGPU/sdiv64.ll
    M llvm/test/CodeGen/AMDGPU/srem64.ll
    M llvm/test/CodeGen/AMDGPU/udiv64.ll
    M llvm/test/CodeGen/AMDGPU/unstructured-cfg-def-use-issue.ll
    M llvm/test/CodeGen/AMDGPU/urem64.ll
    M llvm/test/CodeGen/AMDGPU/vgpr-descriptor-waterfall-loop-idom-update.ll
    M llvm/test/CodeGen/AMDGPU/wave32.ll

  Log Message:
  -----------
  [AMDGPU] MachineLICM cannot hoist VALU

MachineLoop::isLoopInvariant() returns false for all VALU
because of the exec use. Check TII::isIgnorableUse() to
allow hoisting.

That unfortunately results in higher register consumption
since MachineLICM does not adequately estimate pressure.
Therefor I think it shall only be enabled after D107677 even
though it does not depend on it.

Differential Revision: https://reviews.llvm.org/D107859