[PATCH] D20568: [X86][SSE] Replace (V)CVTDQ2PD(Y) and (V)CVTPS2PD(Y) lossless conversion intrinsics with generic IR

[Michael Kuperstein via llvm-commits]

mkuper accepted this revision.
mkuper added a comment.
This revision is now accepted and ready to land.

LGTM with a small nit.


================
Comment at: lib/IR/AutoUpgrade.cpp:416
@@ +415,3 @@
+      if (NumDstElts < SrcTy->getNumElements()) {
+        SmallVector<int, 8> ShuffleMask;
+        for (int i = 0; i != (int)NumDstElts; ++i)
----------------
Shouldn't the 8 here be 2? I think we only get into this case of NumDstElts == 2 (and SrcTy->getNumElements() == 4, which it always is).


Repository:
  rL LLVM

http://reviews.llvm.org/D20568