SLPVectorizer: start constructing chains at stores that are not power of two.
authorNadav Rotem <nrotem@apple.com>
Wed, 19 Jun 2013 15:57:29 +0000 (15:57 +0000)
committerNadav Rotem <nrotem@apple.com>
Wed, 19 Jun 2013 15:57:29 +0000 (15:57 +0000)
commit86e848c849e21f0d0009aeadbe71a8ab6b4e9ef2
treee57b63cb696ace5749820cfde9ed6b1a3813c1f6
parent18e8cf433528f103e7e1022b2b7a6b92bc037c41
SLPVectorizer: start constructing chains at stores that are not power of two.
The type <3 x i8> is a common in graphics and we want to be able to vectorize it.

This changes accelerates bullet by 12% and 471_omnetpp by 5%.

llvm-svn: 184317
llvm/lib/Transforms/Vectorize/VecUtils.cpp
llvm/test/Transforms/SLPVectorizer/X86/odd_store.ll [new file with mode: 0644]