[llvm] [AMDGPU] Align loop headers to prevent instruction fetch split on GFX950 (PR #181999)

[via llvm-commits]

================
@@ -576,6 +576,8 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
 
   bool hasCvtScaleForwardingHazard() const { return HasGFX950Insts; }
 
+  bool hasLoopHeadInstSplitSensitivity() const { return HasGFX950Insts; }
----------------
michaelselehov wrote:

Done

https://github.com/llvm/llvm-project/pull/181999