[PATCH] D141602: [TTI][AArch64] Cost model insertelement and indexed LD1 instructions

[Dave Green via Phabricator via llvm-commits]

dmgreen added a comment.

I have in the past considered the cost from getVectorInstrCost to be less about the exact cost of moving into vector lanes, and more about how much sillyness you are willing to put up with from the SLP vectorizer.

If you have test cases that could be added for the SLP vectorizer, that would be useful too.



================
Comment at: llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp:2127
+    if (I && dyn_cast<LoadInst>(I->getOperand(1)))
+      return 4;
+
----------------
It might be worth using `ST->getVectorInsertExtractBaseCost() + 1`


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D141602/new/

https://reviews.llvm.org/D141602