Path: utzoo!attcan!telly!lethe!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!julius.cs.uiuc.edu!news.cs.indiana.edu!ariel.unm.edu!news From: john@ghostwheel.unm.edu (John Prentice) Newsgroups: comp.lang.misc Subject: Re: f2c experiences Message-ID: <1990Dec7.055715.29344@ariel.unm.edu> Date: 7 Dec 90 05:57:15 GMT Sender: John K. Prentice Organization: University of New Mexico Math Dept., Albuquerque, NM Lines: 54 ----------------------------------------------------------------------- Note: The original posting I am replying to was in comp.lang.fortran. I have posted this reply in both that and comp.lang.misc because I have been vocal in questioning f2c and I wanted to be sure this reached the people involved in the debate. - JP ------------------------------------------------------------------------ In article bglenden@mandrill.cv.nrao.edu (Brian Glendenning) writes: > >Anyway, I decided to run our package through f2c. Here's a little note >I wrote up about my experience. > talks about the converted code and then... > >The resulting f2c code passed the verification suite with flying >colours. This surprised me a bit since I thought that we might run >into parentheses grouping problems since Sun cc is a K&R compiler and >I didn't specify the flags to force f2c to follow Fortran evaluation. > and some more things.... > >These numbers are typical of what we find when bringing up AIPS on any >new system. > some timing comparisons showing the converted C code pretty code to the Fortran speeds... > >What can we conclude from this? Well, the obvious thing is that it >works and we can make AIPS run on machines without Fortran compilers. >Next, I believe that the above performance numbers could be increased >with modest amounts of effort, even with changes as trivial as >compiler command line options. If this is true it may have important >consequences in how we direct "new AIPS." I think we should consider >pursuing this experiment on more interesting machines such as the >Convex and the IBM workstation (or even with gcc on Suns). > I found this quite interesting, particularly since I have been so vocal in questioning f2c. I have heard from several other people about their experiences also. Without making an pretense to a full summary, let me say that so far everyone I have solicited or who has contacted me has had good experiences with f2c. It sounds like a solid piece of coding which is reliable. I haven't heard from anyone who has experiences with problems in converting math packages, though I still can't but wonder if preserving numerical properties isn't asking for an awful lot out of any automatic conversion. In any case, my skeptism about f2c is largely going away, even for scientific code. It appears to be a very impressive code that has been well tested. John Prentice john@unmfys.unm.edu