[llvm] [AArch64][SVE] Use SVE for scalar FP converts in streaming[-compatible] functions (1/n) (PR #118505)

[Paul Walker via llvm-commits]

================
@@ -18989,13 +18989,75 @@ static SDValue performVectorCompareAndMaskUnaryOpCombine(SDNode *N,
   return SDValue();
 }
 
+/// Tries to replace scalar FP <-> INT conversions with SVE in streaming
+/// functions, this can help to reduce the number of fmovs to/from GPRs.
+static SDValue
+tryToReplaceScalarFPConversionWithSVE(SDNode *N, SelectionDAG &DAG,
+                                      const AArch64Subtarget *Subtarget) {
----------------
paulwalker-arm wrote:

This does not feel like the sort of transformation we should be doing early, but rather a post operation legalisation combine.

https://github.com/llvm/llvm-project/pull/118505