[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gcc-fortran
Subject:    Re: gas_dyn regression?
From:       Uros Bizjak <ubizjak () gmail ! com>
Date:       2008-03-24 15:40:30
Message-ID: 47E7CB6E.9080608 () gmail ! com
[Download RAW message or body]

Hello!

> On Sun, 2008-03-23 at 13:00 -0400, Jack Howarth wrote:
> > Looking at the recent results of the nightly polyhedron benchmark
> > runs (http://physik.fu-berlin.de/~tburnus/gcc-trunk/benchmark/ \
> > <http://physik.fu-berlin.de/%7Etburnus/gcc-trunk/benchmark/>), I noticed that \
> > there seems to have been a serious performance regression in the gas-dyn runtime \
> > around late January for the 32-bit gfortran runs. Any ideas about what caused \
> > that?
> 
> I can't find any obvious culprits in the ChangeLogs.
> 
> Can you open a PR?
> 

This is similar, if not the same as "PR 31897, [4.3/4.4 Regression] 30% 
speed regression with -m32 on Opteron with rnflow". Unfortunatelly, this 
problem is AMD specific, I don't see any of these two regressions on Core2.


[1] http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31897

Uros.


[prev in list] [next in list] [prev in thread] [next in thread]