|
From: | Daniel Kraft |
Subject: | Re: Fortran flags |
Date: | Wed, 11 May 2011 11:06:50 +0200 |
User-agent: | Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.9.2.15) Gecko/20110303 Lightning/1.0b2 Thunderbird/3.1.9 |
On 05/11/11 02:16, Jordi Gutiérrez Hermoso wrote:
On 10 May 2011 15:08, Dmitri A. Sergatskov<address@hidden> wrote:2011/5/10 Thomas Weber<address@hidden>:On Tue, May 10, 2011 at 11:20:45AM -0500, Jordi Gutiérrez Hermoso wrote:On 10 May 2011 10:12, Daniel Kraft<address@hidden> wrote:Usually, -O2 should be a good thing for gfortran. And in general, please report any "problems" -O2 causes as gfortran bug.I was wondering why it's not enabled in the default Fortran flags. Instead we merely use -O. I thought the option enabled other optimisation flags that caused problems. Red Hat for example broke ABI compatibility by compiling with -O2: https://bugzilla.redhat.com/show_bug.cgi?id=478856Maybe I don't get it, but I don't see anything in that bug report that even says someting about optimization flags? ThomasI suspect he meant this bug: https://bugzilla.redhat.com/show_bug.cgi?id=520518#c0 (and few ones related) which showed up in all other distributions as well... (It was not really RedHat specific).Yes, thank you, about compilation Fortran flags. I read the first report too quickly. Daniel, this was Dmitri's original report on the matter: http://octave.1599824.n4.nabble.com/help-with-compiling-octave-3-3-51-tp2272804p2282648.html Do you think this is a gcc bug?
I did not look at it in detail, but gfortran 4.1 is already quite old; and I think, at this stage gfortran just started to being somewhat useful. So I strongly suppose that this bug (if it is one) has been fixed long since.
So, my opinion is that -O2 should be really fine (and should help with performance at least a little) -- unless some new problems explicitely come up.
Yours, Daniel -- http://www.pro-vegan.info/ -- Done: Arc-Bar-Cav-Kni-Ran-Rog-Sam-Tou-Val-Wiz To go: Hea-Mon-Pri
[Prev in Thread] | Current Thread | [Next in Thread] |