[llvm] [AMDGPU] Add waterfall intrinsics (PR #192409)

[Nicolai Hähnle via llvm-commits]

================
@@ -1736,6 +1738,7 @@ void GCNPassConfig::addFastRegAlloc() {
 }
 
 void GCNPassConfig::addPreRegAlloc() {
+  addPass(createAMDGPUInsertWaterfallPass());
----------------
nhaehnle wrote:

What's the rationale for inserting the pass here?

https://github.com/llvm/llvm-project/pull/192409