This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug tree-optimization/72517] [7 Regression] 436.cactusADM: More than 40% regression in O3 and Ofast on AMD bdver4 m/c.
- 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: Thu, 28 Jul 2016 08:45:52 +0000
- Subject: [Bug tree-optimization/72517] [7 Regression] 436.cactusADM: More than 40% regression in O3 and Ofast on AMD bdver4 m/c.
- Auto-submitted: auto-generated
- References: <bug-72517-4@http.gcc.gnu.org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=72517
--- Comment #13 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to rguenther@suse.de from comment #12)
> On July 27, 2016 7:02:07 PM GMT+02:00, "Amit.Pawar at amd dot com"
> <gcc-bugzilla@gcc.gnu.org> wrote:
> >https://gcc.gnu.org/bugzilla/show_bug.cgi?id=72517
> >
> >--- Comment #11 from Amit Pawar <Amit.Pawar at amd dot com> ---
> >Improvement is seen but not same as best one.
> >
> >
> >Current scores are
> >Flags Latest trunk
> >O3 :31.4
> >O3 -march=bdver4 :31.1
> >Ofast :32.1
> >Ofast -march=bdver4 :31.2
> >
> >but good one.
> >Flags Good Trunk
> >O3 37.69
> >O3 bdver4 36.55
> >Ofast 40.36
> >Ofast bdver4 38.31
> >
> >Can you please check at your end?
>
> I checked r237473 plus the patch against r237472 and the regression was
> fully fixed. I did not yet check current trunk but if there is any
> regression it is a new one caused by sth else. Bisection with the fix for
> this regression might tell.
Double-checking against trunk, thus base is r237472 and peak is r238807 I get
436.cactusADM 11950 320 37.3 * 11950 318 37.5 *
for -Ofast -march=native (three-run result). This is FAM 21 Model 96 Stepping
1 detected as bdver4 by GCC.