[llvm] 1f15e39 - AMDGPU/GlobalISel: Don't pointlessly check for convergent intrinsics

[Matt Arsenault via llvm-commits]

Author: Matt Arsenault
Date: 2023-09-15T23:32:19+03:00
New Revision: 1f15e39d8194885cb0aea618cce12eb416b2129f

URL: https://github.com/llvm/llvm-project/commit/1f15e39d8194885cb0aea618cce12eb416b2129f
DIFF: https://github.com/llvm/llvm-project/commit/1f15e39d8194885cb0aea618cce12eb416b2129f.diff

LOG: AMDGPU/GlobalISel: Don't pointlessly check for convergent intrinsics

The set of handled intrinsics for fneg combines aren't convergent. The only
case we might want to handle is mov_dpp.

Added: 
    

Modified: 
    llvm/lib/Target/AMDGPU/AMDGPUCombinerHelper.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/AMDGPU/AMDGPUCombinerHelper.cpp b/llvm/lib/Target/AMDGPU/AMDGPUCombinerHelper.cpp
index d4f38202fc3c50d..d48916670112c0f 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUCombinerHelper.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUCombinerHelper.cpp
@@ -42,8 +42,7 @@ static bool fnegFoldsIntoMI(const MachineInstr &MI) {
   case AMDGPU::G_AMDGPU_FMIN_LEGACY:
   case AMDGPU::G_AMDGPU_FMAX_LEGACY:
     return true;
-  case AMDGPU::G_INTRINSIC:
-  case AMDGPU::G_INTRINSIC_CONVERGENT: {
+  case AMDGPU::G_INTRINSIC: {
     unsigned IntrinsicID = cast<GIntrinsic>(MI).getIntrinsicID();
     switch (IntrinsicID) {
     case Intrinsic::amdgcn_rcp: