This is the mail archive of the
mailing list for the GCC project.
Re: gcc compile-time performance
- From: Tim Prince <tprince at computer dot org>
- To: Toon Moene <toon at moene dot indiv dot nluug dot nl>,Robert Dewar <dewar at gnat dot com>
- Cc: gdr at codesourcery dot com, gcc at gcc dot gnu dot org, scott at coyotegulch dot com
- Date: Sat, 18 May 2002 15:07:28 -0700
- Subject: Re: gcc compile-time performance
- References: <20020518200706.28165F28EB@nile.gnat.com> <3CE6B6CC.4A8AD401@moene.indiv.nluug.nl>
- Reply-to: tprince at computer dot org
On Saturday 18 May 2002 13:17, Toon Moene wrote:
> Robert Dewar wrote:
> > <<I can put this stronger: The very reason our numerical weather
> > prediction software is still written in Fortran 77 is due to the fact
> > that that way we can be assured of the collaboration of [the Weather
> > Institutes] of smaller European countries, where Universities can't
> > afford Real Unix Workstations.
> > What exactly is a "real unix work station". Given that the best cost
> > performance on small scale unix platforms now comes from running
> > GNU/Linux on a cheap PC, I find that hard to understand.
> Well, duh ! The point is that noone trusts the proprietary compilers -
> even if they're `free' (i.e., gratis).
> People want a free (according to our - GNU - sense of free) supported
> Fortran 90/95 compiler before they switch.
Several organizations I have dealt with continue to maintain mostly
g77-compatible source code, using only a few selected f90 features, if any,
due to uncertainties in quality of implementation across multiple platforms,
and the priority on portability. At this time, gcc/g77 is, AFAIK, one of
only 2 compiler families which support ia32 sse/sse2 code generation.