Return to BSD News archive
Xref: sserve comp.os.386bsd.bugs:1541 comp.os.386bsd.questions:5580
Newsgroups: comp.os.386bsd.bugs,comp.os.386bsd.questions
Path: sserve!newshost.anu.edu.au!munnari.oz.au!cs.mu.OZ.AU!summer
From: summer@ee.mu.OZ.AU (Mark Summerfield)
Subject: Alt. Mathlib "Floating point exception"
Message-ID: <9327712.6347@mulga.cs.mu.OZ.AU>
Sender: news@cs.mu.OZ.AU
Organization: Dept of Electrical & Electronic Engineering, University of Melbourne
Date: Mon, 4 Oct 1993 01:59:59 GMT
Lines: 21
I have been using Glenn Geers' alternative maths library, fpulib (ported
by Risner@ms.ukyedu). Usually it is good -- more accurate and faster
than the standard libm. However, on some programs it causes floating
point exceptions which seem unrelated to the code in which they occur
(they're not "real" overflows, or divisions by zero or anything). The
same problems do not occur using the standard libm.
There seems to be little correlation between the specifics of the
program and the exceptions -- sometimes a small change will cause a
working program to fail. In the most recent case, the only calls
the program was using (other than adds, multiplies etc.) were exponential
and sine.
Has anyone else seen this behaviour and/or know what a fix might be?
Mark.
--------------------------------------------------------
Mark Summerfield, Photonics Research Laboratory
Department of Electrical and Electronic Engineering, University of Melbourne
ACSnet[AARN/Internet]: summer@ee.mu.oz[.au]
--------------------------------------------------------