This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Compilation flags in libgfortran
- From: Igor Zamyatin <izamyatin at gmail dot com>
- To: Kyrill Tkachov <kyrylo dot tkachov at arm dot com>
- Cc: "pinskia at gmail dot com" <pinskia at gmail dot com>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>, Yuri Rumyantsev <ysrumyan at gmail dot com>
- Date: Wed, 16 Oct 2013 15:58:25 +0400
- Subject: Re: Compilation flags in libgfortran
- Authentication-results: sourceware.org; auth=none
- References: <CAKdSQZk4gBwLOXPNL+Z+Nc8476S0acuFMKYS3QLG=hkzzLCqbQ at mail dot gmail dot com> <D1AD949E-E6A9-44F3-BE2F-E850B53139BF at gmail dot com> <525E68F5 dot 1080706 at arm dot com>
Yeah, this is my point exactly. Atom case seems just triggered that fact.
On Wed, Oct 16, 2013 at 2:22 PM, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote:
> On 16/10/13 10:37, pinskia@gmail.com wrote:
>>>
>>> On Oct 15, 2013, at 6:58 AM, Igor Zamyatin <izamyatin@gmail.com> wrote:
>>> Hi All!
>>>
>>> Is there any particular reason that matmul* modules from libgfortran
>>> are compiled with -O2 -ftree-vectorize?
>>>
>>> I see some regressions on Atom processor after r202980
>>> (http://gcc.gnu.org/ml/gcc-cvs/2013-09/msg00846.html)
>>>
>>> Why not just use O3 for those modules?
>>
>> -O3 and -O2 -ftree-vectorize won't give much performance difference. What
>> you are seeing is the cost model needs improvement; at least for atom.
>
> Hi all,
> I think http://gcc.gnu.org/ml/gcc-patches/2013-09/msg01908.html introduced
> the new "cheap" vectoriser cost model that favors compilation time over
> runtime performance and is set as default for -O2. -O3 uses the "dynamic"
> model which potentially gives better runtime performance in exchange for
> longer compile times (if I understand the new rules correctly).
> Therefore, I'd expect -O3 to give a better vector performance than -O2...
>
> Kyrill
>
>