Re: fortran compilier recommendations
I have used the Absoft compiler because it supported Vax style fortran structures which we needed for some code we were porting. It is OK as a compiler. I needed it to work with g77 since we used several libraries built with g77 which proved to be very difficult to compile with ABSOFT due to many workarounds to get them to work with g77. I succeeded with significant work in linking and correctly running a program that used Absoft f77, g77, gcc and g++. I consider that to be to ABSOFT's credit. I did find one compiler bug involving character functions, and I had trouble with debugging symbols not being properly included if the object file was sent to a different directory from the source file. I have not investigated optimization significantly. It turns out the g++ piece is the slowpoke and that is not my responsibility. If I could survive with g77 I would do so. ABSOFT is not that great advantage over g77 in my case except for the fortran structures. Have tried the comp.lang.fortran newsgroup. They have significant discussions of issues like this. Mike __ Michael Procario EMAIL: [EMAIL PROTECTED] Department of Physics PHONE: 412/268-3887 Carnegie Mellon University Another casualty of applied metaphysics __
Re: fortran compilier recommendations
On Wed, Mar 17, 1999 at 08:20:16PM -0500, Michael Procario wrote: is the slowpoke and that is not my responsibility. If I could survive with g77 I would do so. ABSOFT is not that great advantage over g77 in my case except for the fortran structures. I'm using the Portland Group F90 and HPF compilers quite happily. They don't seem to offer that big a win over g77 or f2c other than the Fortran 90 support (and autoparallelisation for those occasions where it works) - I probably wouldn't bother at all if I'd had to pay for them. We seem to be getting free spreadsheets, so now the only thing I used that I can't get as free software is a F90 compiler, but sadly few people even seem interested in helping improve g77. :-( Have tried the comp.lang.fortran newsgroup. They have significant discussions of issues like this. The best summary of what's out there for Linux is probably the page put together by Jeff Templon at http://studbolt.physast.uga.edu/templon/fortran.html -- Mark Brown mailto:[EMAIL PROTECTED] (Trying to avoid grumpiness) http://www.tardis.ed.ac.uk/~broonie/ EUFShttp://www.eusa.ed.ac.uk/societies/filmsoc/ pgpkqPAC9L6YH.pgp Description: PGP signature
Re: fortran compilier recommendations
On Wed, 10 Mar 1999 [EMAIL PROTECTED] wrote: We have a large application currently running on Dec Alpha 500/500 machines which we are investigating porting to a Linux platform. It consist of a SmallTalk engine that uses C and Fortran routines to do it's actual work. While all the Fortran compiles with g77/egcs we have some issues with optimization greater than -O0. This code is legacy code originally written on VMS and earlier platforms. We have tried the Portland Group's compiler with only slightly more success. So finally the question: Does any one have a recommendation for a third party compiler for Fortran 77 that has been throughly rung out by the user community? Hi. Are you porting to x86 or Alpha? I've been so happy with eg77 (Debian's g77) on PPro and PII machines that I was able to avoid trying those other choices (Portland, Absoft, etc.). The optimization up to level 3 (-O3) has worked just fine on the Intel platform in my experience, and most f77 programs port without any changes. Of course, Digital's f77 with the -O4 switch produces faster code per MHz on our 600MHz alpha runing DUNIX. Anyway, sorry I can't be of any help, but I'll be following this thread to see if anything interesting comes up. Thanks. Syrus. -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Syrus Nemat-Nasser [EMAIL PROTECTED]UCSD Physics Dept.
fortran compilier recommendations
Hi All, We have a large application currently running on Dec Alpha 500/500 machines which we are investigating porting to a Linux platform. It consist of a SmallTalk engine that uses C and Fortran routines to do it's actual work. While all the Fortran compiles with g77/egcs we have some issues with optimization greater than -O0. This code is legacy code originally written on VMS and earlier platforms. We have tried the Portland Group's compiler with only slightly more success. So finally the question: Does any one have a recommendation for a third party compiler for Fortran 77 that has been throughly rung out by the user community? Thanks, Jim - Jim Harsh E-Mail: [EMAIL PROTECTED] Los Alamos National Laboratory, MS P940 Los Alamos, NM 87545 505-665-0485, FAX - 505-665-3359 Date: 10-Mar-99,Time: 13:08:22 -
