[llvm] AMDGPU: Custom lower fptrunc vectors for f32 -> f16 (PR #141883)

[Shilei Tian via llvm-commits]

================
@@ -145,6 +145,7 @@ class SITargetLowering final : public AMDGPUTargetLowering {
 
   /// Custom lowering for ISD::FP_ROUND for MVT::f16.
   SDValue lowerFP_ROUND(SDValue Op, SelectionDAG &DAG) const;
+  SDValue splitFP_ROUNDVectorOp(SDValue Op, SelectionDAG &DAG) const;
----------------
shiltian wrote:

Did I miss anything here? This function was added but not implemented?

https://github.com/llvm/llvm-project/pull/141883