[llvm-branch-commits] [llvm] [AMDGPU] Add structural stall heuristic to scheduling strategies (PR #169617)

[Austin Kerbow via llvm-branch-commits]

================
@@ -313,6 +313,10 @@ unsigned GCNHazardRecognizer::PreEmitNoops(MachineInstr *MI) {
   return std::max(W, NopPadding.getValue());
 }
 
+unsigned GCNHazardRecognizer::getHazardWaitStates(MachineInstr *MI) const {
+  return const_cast<GCNHazardRecognizer *>(this)->PreEmitNoopsCommon(MI);
----------------
kerbowa wrote:

There will be no hazardrec instantiated preRA so it will not have an impact.

https://github.com/llvm/llvm-project/pull/169617