[llvm] [X86] Lower bf16->f32/f64 fpext without a GPR round-trip (PR #218359)

[Simon Pilgrim via llvm-commits]

================
@@ -22973,6 +22975,38 @@ static SDValue LowerFP_TO_FP16(SDValue Op, SelectionDAG &DAG) {
   return Res;
 }
 
+SDValue X86TargetLowering::LowerBF16_TO_FP(SDValue Op,
+                                           SelectionDAG &DAG) const {
+  SDLoc DL(Op);
+  SDValue Src = Op.getOperand(0);
+  // Operand is usually already softened to i16 by type legalization.
+  if (Src.getValueType() == MVT::bf16)
----------------
RKSimon wrote:

you can remove the if check - getBitcast is going to do it anyhow

https://github.com/llvm/llvm-project/pull/218359