[llvm] [AMDGPU] Improve AMDGPU sqrt and inverse sqrt handling for bf16 (PR #180291)

[Matt Arsenault via llvm-commits]

================
@@ -13184,6 +13187,45 @@ SDValue SITargetLowering::lowerFSQRTF64(SDValue Op, SelectionDAG &DAG) const {
                      Flags);
 }
 
+SDValue SITargetLowering::lowerFSQRTBF16(SDValue Op, SelectionDAG &DAG) const {
----------------
arsenm wrote:

This is mostly duplicating the f32 path. Can you merge these into one function? 

https://github.com/llvm/llvm-project/pull/180291