http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54717
--- Comment #6 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-09-27 09:28:04 UTC --- (In reply to comment #4) > The slowdown is mostly hidden by -fno-tree-loop-if-convert. I would say this means we have more vectorization opportunities after the patch. Opportunities that might end up being not profitable. Sergey, are those differences you quote the only differences?