This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH][i386] Enable vector_loop in memset expanding and merge expanders for memset and memmov
- From: Jan Hubicka <hubicka at ucw dot cz>
- To: "Michael V. Zolotukhin" <michael dot v dot zolotukhin at gmail dot com>
- Cc: Jan Hubicka <hubicka at ucw dot cz>, Jan Hubicka <jh at suse dot cz>, Uros Bizjak <ubizjak at gmail dot com>, gcc <gcc-patches at gcc dot gnu dot org>
- Date: Mon, 21 Oct 2013 15:36:38 +0200
- Subject: Re: [PATCH][i386] Enable vector_loop in memset expanding and merge expanders for memset and memmov
- Authentication-results: sourceware.org; auth=none
- References: <20130930122132 dot GB3461 at msticlxl57 dot ims dot intel dot com> <20131010162249 dot GA24619 at atrey dot karlin dot mff dot cuni dot cz> <20131020092018 dot GC3461 at msticlxl57 dot ims dot intel dot com> <20131020093649 dot GA18453 at atrey dot karlin dot mff dot cuni dot cz> <20131021085901 dot GD3461 at msticlxl57 dot ims dot intel dot com>
> > The patch is OK.
> Thanks, the patch was committed.
> > > That's a good point. I added a check for this case - so if CONST_INT is passed
> > > we assume that mode is QI. But usually promoted value resides in a register, so
> > > we don't go over-conservative here.
> > Hmm, so if we use broadcast constant, then we won't end up having CONST_INT here?
> > It is OK then.
> Currently we don't end up with CONST_INT. But maybe we'd need to revisit this
> place later.
OK, I merged in my misaligned prologues changes and will post patch after full
testing. It seemed to go seamlessly. I spotted there are still few places for
cleanup, so i will try to handle there incrementally.
Do you have plans on enabling the SSE memcpy/memset on Atom now? What about
the move_by_pieces/set_by_pieces via SSE now?
> > Honza