--- Comment #4 from ramiro86 at hotmail dot com 2009-04-21 00:10 ---
I've attached a simple testcase. The system I'm running this on is a q6600 with
64-bit Linux.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39821
--- Comment #3 from ramiro86 at hotmail dot com 2009-04-21 00:08 ---
Created an attachment (id=17660)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17660&action=view)
tarball of a simple testcase
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39821
--- Comment #2 from ubizjak at gmail dot com 2009-04-20 20:52 ---
(In reply to comment #1)
> but the widening unpacking results in absymal code generated. Where are
> all the shifts coming from?
Not from unpacking, but from mulv2di pattern from sse.md
Can you please attach full sourc
--- Comment #1 from rguenth at gcc dot gnu dot org 2009-04-20 09:26 ---
The vectorizer creates
vect_var_.128_46 = M*vect_p.123_44{misalignment: 0};
vect_var_.129_47 = [vec_unpack_lo_expr] vect_var_.128_46;
vect_var_.129_48 = [vec_unpack_hi_expr] vect_var_.128_46;
vect_var_.135_5
