[llvm] [AArch64] Account for fixed-SVE high-lane insertelement/extractelement costs (PR #219244)

[Utpal Bora via llvm-commits]

================
@@ -4599,6 +4599,21 @@ InstructionCost AArch64TTIImpl::getVectorInstrCostHelper(
     if (Index == 0 && !Ty->getScalarType()->isIntegerTy())
       return 0;
 
+    // SVE has no scalar move to or from an arbitrary lane above the low
+    // 128-bit portion of a Z register, e.g. there is no equivalent of
+    // "mov z0.d[9], d0". Fixed-length vectors wider than 128 bits therefore
+    // need extra work when scalarizing inserts/extracts for those lanes, so
+    // model them as more expensive than ordinary NEON lane accesses.
+    if (ST->useSVEForFixedLengthVectors() && isa<FixedVectorType>(Ty) &&
+        Ty->getScalarType()->isFloatingPointTy()) {
+      unsigned LowElts = 128 / Ty->getScalarSizeInBits();
----------------
utpalbora wrote:

Done

https://github.com/llvm/llvm-project/pull/219244