[all-commits] [llvm/llvm-project] 718aec: [AMDGPU] Improve v_cmpx usage on GFX10.3.

[Thomas Symalla via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 718aec209c891487294d8a6199cf12c796c6e901
      https://github.com/llvm/llvm-project/commit/718aec209c891487294d8a6199cf12c796c6e901
  Author: Thomas Symalla <thomas.symalla at amd.com>
  Date:   2022-03-25 (Fri, 25 Mar 2022)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
    M llvm/lib/Target/AMDGPU/SIInstrInfo.h
    M llvm/lib/Target/AMDGPU/SIInstrInfo.td
    M llvm/lib/Target/AMDGPU/SIOptimizeExecMasking.cpp
    M llvm/lib/Target/AMDGPU/SIShrinkInstructions.cpp
    M llvm/lib/Target/AMDGPU/VOPCInstructions.td
    M llvm/test/CodeGen/AMDGPU/branch-relaxation-gfx10-branch-offset-bug.ll
    A llvm/test/CodeGen/AMDGPU/vcmp-saveexec-to-vcmpx.ll
    A llvm/test/CodeGen/AMDGPU/vcmp-saveexec-to-vcmpx.mir
    M llvm/test/CodeGen/AMDGPU/wqm.ll

  Log Message:
  -----------
  [AMDGPU] Improve v_cmpx usage on GFX10.3.

On GFX10.3 targets, the following instruction sequence

v_cmp_* SGPR, ...
s_and_saveexec ..., SGPR

leads to a fairly long stall caused by a VALU write to a SGPR and having the
following SALU wait for the SGPR.

An equivalent sequence is to save the exec mask manually instead of letting
s_and_saveexec do the work and use a v_cmpx instruction instead to do the
comparison.

This patch modifies the SIOptimizeExecMasking pass as this is the last position
where s_and_saveexec instructions are inserted. It does the transformation by
trying to find the pattern, extracting the operands and generating the new
instruction sequence.

It also changes some existing lit tests and introduces a few new tests to show
the changed behavior on GFX10.3 targets.

Same as D119696 including a buildbot and MIR test fix.

Reviewed By: critson

Differential Revision: https://reviews.llvm.org/D122332