Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!usc!elroy.jpl.nasa.gov!jarthur!nntp-server.caltech.edu!roo!bchen
From: bchen@roo.caltech.edu (Bing-Qing Chen)
Newsgroups: comp.sys.next
Subject: Abosoft Fortran
Keywords: Absoft, Fortran, f2c
Message-ID: <1991Mar10.234219.5210@nntp-server.caltech.edu>
Date: 10 Mar 91 23:42:19 GMT
Sender: news@nntp-server.caltech.edu
Organization: California Institute of Technology, Pasadena
Lines: 14

Does anyone else notice that Abosoft Fortran usually will not give the
correct results? Or am I the only one who is being so unlucky that the
only 2 Fortran program I have tried to compile with Absoft Fortran gave
wrong results? One of the program is just a direct port from a C program,
the results I got from all other Fortran compilers and f2c were identical
to the C result. But the Abosoft result is off by nearly 20%! The other 
program is a VMS Fortran program. This one even gave different results on
different processors (030 next and 040 next), but neither of them are
correct.  Anyway, if you use Abosoft Fortran, be sure to check the result
with f2c or other Fortran compilers. In fact, f2c is not only faster and
free it will even give the right result! 

- Bing Chen
bchen@pooh.caltech.edu