This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug tree-optimization/71271] SLP loop vectorized twice
- From: "rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Wed, 25 May 2016 08:32:08 +0000
- Subject: [Bug tree-optimization/71271] SLP loop vectorized twice
- Auto-submitted: auto-generated
- References: <bug-71271-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71271
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |missed-optimization
Status|UNCONFIRMED |NEW
Last reconfirmed| |2016-05-25
Ever confirmed|0 |1
--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Heh. I suppose the vectorizer could annotate the unvectorized loop copy with
some flag. But OTOH BB vectorization might end up partially vectorizing the
loop - it sounds hard to detect the case where it would be equivalent to
full loop vectorization.
So, confirmed. But as missed loop vectorization / unnecessary versioning for
aliasing. It's perfectly fine to SLP vectorize this w/o alias versioning.