Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!usc!apple!agate!violet.berkeley.edu!ilan343
From: ilan343@violet.berkeley.edu (Geraldo Veiga)
Newsgroups: comp.unix.sysv386
Subject: Re: WANTED:  Best FORTRAN compiler for SCO UNIX/386 3.2.2 ODT 1.1
Message-ID: <1991Apr1.083124.12263@agate.berkeley.edu>
Date: 1 Apr 91 08:31:24 GMT
References: <31502@usc>
Sender: usenet@agate.berkeley.edu (USENET Administrator)
Organization: University of California, Berkeley
Lines: 22

In article <31502@usc> demikhov@neuro.usc.edu (V P Demikhov) writes:
>We are converting a large (300,000+ line) system of FORTRAN & C routines
>comprising an image processing system donated in source code form to our
>laboratory from a large national science agency.  Can anyone give advice
>about which compiler & vendor is the most reliable, bug free, capable of
>generating code for large data spaces and large code spaces?

I am also interested in hearing about FORTRAN compilers for 386 Unix.

I have been programming mostly in FORTRAN on my 486.  My setup for the
moment is f2c plus cc or gcc.  This combination has been working
perfectly so far, as long as you limit yourself to very vanilla f77
(no vax extensions etc, which is fine with me for portability sake).
Using f2c avoids a few headaches when you mix FORTRAN and C, and
allows the use of standard UNIX debuggers.

I had an opportunity of trying out an old version of the NDP/386 FORTRAN
compiler and f2c produced faster code for the Linpak and Whetstone
benchmarks.

For those not aware of it, f2c is the FORTRAN to C translator 
available through the netlib.