[PATCH] D78216: [TTI] Add DemandedElts to getScalarizationOverhead

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/lib/Target/X86/X86TargetTransformInfo.cpp:2610
+        (MScalarTy == MVT::f32 && ST->hasSSE41())) {
+      Cost += BaseT::getScalarizationOverhead(Ty, DemandedElts, Insert, false);
+    } else if (LT.second.isVector()) {
----------------
Doesn't this over count the cost of inserting multiple elements into the upper half of a 256 bit vector? We'll cost a subvector insert for each element, but we should be able to share.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D78216/new/

https://reviews.llvm.org/D78216