Date: Mon, 01 May 2000 15:03:00 -0700 From: Kent Stewart <kstewart@3-cities.com> To: Rahul Siddharthan <rsidd@physics.iisc.ernet.in> Cc: freebsd-questions@freebsd.org Subject: Re: Fortran compilers -- any comparisons/recommendations? Message-ID: <390DFF14.914A8C68@3-cities.com> References: <20000502021251.B3832@physics.iisc.ernet.in>
next in thread | previous in thread | raw e-mail | index | archive | help
Rahul Siddharthan wrote: > > Does anyone have experience with Fortran compilers other than > g77, eg NAG etc? Do linux compilers work? I imagine that > at least static binaries generated on a linux machine should run. > How do commercial compilers compare with g77 (gcc2.95 version), > speed-wise? Any pointers to recent comparisons on the web? This is the part of the *nix's where things go downhill. I have DEC Visual Fortran 5.0 on my NT system and g77 on FreeBSD. There isn't any comparison. The whole DEC system is better. It doesn't win all of the contests because I had a personal project that I wanted to make a monte carlo program from the Cray work on pc's. I actually made it run first on NT (because of Microsoft's Visual debugger) but I couldn't have made it run without access to the g77 and ddd/dbg interface on FreeBSD. The code had some F66 carryovers and msvd would just roll over and die with the memory out of range access. The combination of ddd/dbg left me on the line causing the signal error. I figure finding that line is 80% of debugging a code. The problem was that ddd wouldn't show me the arrays that were being used to create the index and msvd would. I tried adding a diagnostic write and it generated a 70MB output file before it died. Pining it down was going to generate even more output. That was a major plus and minus for both of them. All I had to do was run the mouse pointer over the index on msvd and a value of 109 followed by 6 more numbers popped up. I didn't have that much real*8 memory or disk. Finding the options on each compiler that allowed the f66'ism to work was almost as much work :). I've been looking at Vast-90 or one of the F90 versions. The personal versions are much cheaper but they don't cover what I want to do. So far I just haven't decided to spend the money to get one of their real versions. The visual frontend's such as Code Crusader, Code Forge, Kdevelop, and others are mostly frontends. DEC's Fortran was more tightly coupled and that is better. The worst experience was with Lahey 32/m on an NT system. Their make invoked the Lehey compiler for each Fortran module. After 7 hours, it was only half through compiling the program. We ordered Microsoft Power Fortran, which they later sold to DEC, and added the modules and told it to build. 2 minutes later it was through. Fourteen hours to two minutes is one hell of a ratio for the Microsoft compiler. Dec made it just better. I don't know how much effort Compaq will put into keeping it up to date. Most of the language support is in all of them. How smart they appear depends on the Makefile you create. Most of the programs I have won't allow "g77 -c $(SRC)" because there are simple too many modules and you run out of command line space. Generating the makefile to begin with is another editing problem. How many obscure names can you type without making an error you can't see. That becomes a "how many forehead whacks does it take to cause some brain damage" :). Kent > > Thanks > Rahul. > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-questions" in the body of the message -- Kent Stewart Richland, WA mailto:kstewart@3-cities.com http://www.3-cities.com/~kstewart/index.html FreeBSD News http://daily.daemonnews.org/ SETI(Search for Extraterrestrial Intelligence) @ HOME http://setiathome.ssl.berkeley.edu/ Hunting Archibald Stewart, b 1802 in Ballymena, Antrim Co., NIR http://www.3-cities.com/~kstewart/genealogy/archibald_stewart.html To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?390DFF14.914A8C68>