Date: Wed, 6 Mar 1996 11:35:11 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Erik Luijten <erik AT tntnhb3 DOT tn DOT tudelft DOT nl>
Cc: djgpp <djgpp AT delorie DOT com>
Subject: Re: paranoia
In-Reply-To: <9603051656.AA06324@tntnhb3.tn.tudelft.nl>
Message-Id: <Pine.SUN.3.91.960306113120.25416N-100000@is>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On Tue, 5 Mar 1996, Erik Luijten wrote:

> disturbed me. Therefore I have compiled paranoia (from netlib), to see whether
> this program reveals some problems. Well, I have compiled it without
> optimization (the program completely choked when I tried -O2, why?) 

Any program that tries to reveal numerical intricacies of a machine or a 
library MUST be compiled without optimizations, or else the results will 
depend on the compiler make and version and the optimizations every 
switch enables.  I think the docs for paranoia explicitly tell you to 
disable optimizations.

> paranoia said the following:
> 
> No chopping nor rounding on * / + - sqrt: sticky bit incorrectly used or not
> used at all. 

Was that with or without -lm?  Does the above result depend on whether 
you use -lm?