This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: gnu fortran compiler on 64 bit machines
- From: "Janne Blomqvist" <blomqvist dot janne at gmail dot com>
- To: zuheyr <alsalihi at vki dot ac dot be>
- Cc: fortran at gcc dot gnu dot org
- Date: Tue, 2 Oct 2007 14:19:22 +0300
- Subject: Re: gnu fortran compiler on 64 bit machines
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=AlpwFs9ekcZXyRpmwhecaVYH4OoVWi1oDSSTRpHVuCo=; b=rWEy/NwbiveXYl7q3CzizljIFecBnlRVeCz5pttYkZxpGHU75A6L/++V2EcI/NKEZ56duBWBnuB8DsRbQyiUmAWunBkDVLkOdd6q8KHTZvHhmsy2MHqtpwpIvCxppOQYSD5kzalULzBGkqXTgM7EaDIhO14gasA44eIFUeSdAhg=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=OTxg55R+36WpSWMNVXz3i2WdVhwmdMfr71PRYD7MLstnXJIrnHLvQiN6IBpqd5UiG8NC6INR0qWyKTAspe9MPVkEpICnx5KOOp6vwB/r4a+3dtAnhxd2UhtTFJ+JFFQjK+ODA5avhBHd/LXETd63p0c7S6rg9Inq0Lybo1nyHvg=
- References: <12996731.post@talk.nabble.com>
On 10/2/07, zuheyr <alsalihi@vki.ac.be> wrote:
>
> Hello,
>
> I am using GNU fortran on 32/64 bit Fedora Core 6 Linux machines using
> Pentium IV and AMD
> Athlon processors. I use fortran 77 and fortran 90.
>
> I have seen serious differences between the gnu fortran compiled codes and
> Lahey fortran compiler
> code. I mean not in performance but also in execution as the results
> seriously differ.
> This is for 32 bit machines. The application is Navier-Stokes equation
> solver, so large programs.
>
> However, most notably there are also big differences between gnu fortran
> compiled codes on 32 and 64 bit machines.
> The 64 bit code does not at all run on what 32 bit gnu compiled code runs.
>
> In the compilations I used the simplest -O flag but there are differences
> even with -g flag.
Does your code compile and run properly if you compile with "-Wall
-pedantic-errors -C"? If not, fix your code.
Is your code numerically unstable? I.e. small differences in input
and/or differences in evaluation order/precision between different
compilers and different optimization levels, cause big differences in
output. If so, your results will be more or less random depending of
which compiler you use. Which is then the correct result? Well, noone
knows.
--
Janne Blomqvist