[PATCH] D6818: [SLPVectorization] Vectorize flat addition in a single tree (+(+(+ v1 v2) v3) v4)

[Nadav Rotem via llvm-commits]

nadav added a comment.

It looks like this patch is not ready for review.

Before submitting it again please run this code on the llvm test suite and collect performance numbers (runtime and compile time). I want to make sure we are not regressing and assess the wins.


================
Comment at: lib/Transforms/Vectorize/SLPVectorizer.cpp:79
@@ +78,3 @@
+static bool IsReturn = false;
+
+static bool IsHAdd = false;
----------------
I agree with jmolloy. No globals please. 


Repository:
  rL LLVM

http://reviews.llvm.org/D6818