[llvm] [X86] Combine `uitofp <v x i32> to <v x half>` (PR #121809)

[Phoebe Wang via llvm-commits]

================
@@ -1777,6 +1777,36 @@ void VectorLegalizer::ExpandUINT_TO_FLOAT(SDNode *Node,
   assert((BW == 64 || BW == 32) &&
          "Elements in vector-UINT_TO_FP must be 32 or 64 bits wide");
 
+  // If STRICT_/FMUL is not supported by the target (in case of f16) replace the
+  // UINT_TO_FP with a larger float and round to the smaller type
+  if ((!IsStrict &&
+       (TLI.getOperationAction(ISD::FMUL, DstVT) == TargetLowering::Expand ||
+        TLI.getOperationAction(ISD::FMUL, DstVT) == TargetLowering::Promote)) ||
+      (IsStrict && (TLI.getOperationAction(ISD::STRICT_FMUL, DstVT) ==
+                        TargetLowering::Expand ||
+                    TLI.getOperationAction(ISD::STRICT_FMUL, DstVT) ==
+                        TargetLowering::Promote))) {
----------------
phoebewang wrote:

ditto.

https://github.com/llvm/llvm-project/pull/121809