[llvm] [LoopVectorize][NFC] Cache the result of getVScaleForTuning (PR #124732)

[Florian Hahn via llvm-commits]

================
@@ -1554,9 +1554,32 @@ class LoopVectorizationCostModel {
   /// trivially hoistable.
   bool shouldConsiderInvariant(Value *Op);
 
+  /// Return the value of vscale used for tuning the cost model.
+  std::optional<unsigned> getVScaleForTuning() const { return VScaleForTuning; }
+
 private:
   unsigned NumPredStores = 0;
 
+  std::optional<unsigned> VScaleForTuning;
----------------
fhahn wrote:

Could yo u document the field? 

https://github.com/llvm/llvm-project/pull/124732