Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!news.unimelb.edu.au!munnari.OZ.AU!news.Hawaii.Edu!news.caldera.com!news.eli.net!uunet!in2.uu.net!128.230.129.106!news.maxwell.syr.edu!ais.net!vixen.cso.uiuc.edu!uwm.edu!msunews!ariel.cps.msu.edu!punch From: punch@cps.msu.edu (Bill Punch) Newsgroups: comp.unix.bsd.freebsd.misc Subject: FP exceptions under FreeBSD Date: 13 Jun 1997 15:31:25 GMT Organization: MSU GARAGe Lines: 30 Distribution: world Message-ID: <5nrp4d$obd$1@msunews.cl.msu.edu> NNTP-Posting-Host: ariel.cps.msu.edu Originator: punch@ariel.cps.msu.edu Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:42848 I have a program (called lilgp, a C program for genetic programming) that I have used on a number of unix and pc platforms. For those that don't know, a genetic program generates "programs", actually trees of functions, to perform some operation. When I do this under FreeBSD, and the programs use floating point operations, I typically get FPE core dumps. Debugging shows some pretty standard operations (*, +, etc.), albeit on either some very large or small numbers (xyz**223, or abc**-103, numbers like that). These numbers are generated randomly and so what they are depends on the random seed. The question is this. I find it hard to believe that this is really a FP problem. If I run this program under conditions of little memory usage, it seems to work. If I push it to use more memory, I get FPE everywhere. Could be a memory leak or something, but why would it show up as an FPE? Or is there indeed a problem in FP I should be aware of? I'm running this on a portable 75MHz pentium, 16Mb memory. I've tried various compile flags including -mno-ieee-float (or whatever it is to turn off ieee fp). Any help appreciated. >>>bill punch<<< punch@cps.msu.edu -- Bill Punch