From: Mark W. Spychalla (spy@castlab.uucp)
Date: 12/11/92


From: spy@castlab.uucp (Mark W. Spychalla)
Subject: floating point exceptions
Date: 11 Dec 1992 14:06:10 CST


Hello,

  I have had many problems with the floating point under Linux as well. I have
been running the first SLS distribution (linux 0.96 I believe) with Linux extfs
and X windows for quite some time on a V-com DX2/66 with 8 Meg RAM, 160 Meg
Disk Space, and ET4000 based local bus video card. I have been EXTREMELY
pleased with my Linux performance so far for the animation/rendering
applications I have been developing.
  
I have done the following kludges to work around my float problems:

1) Ignore floating point error signals (!!!) by using

   #include <signal.h>
   ...
   main()
   {
   signal(SIGFPE, SIG_IGN);
   ...
   }

2) Compile with the soft floating point libraries instead of the default
   libraries they seem to be of better quality than the default ones.

   gcc -msoft-float -o foo foo.c -lsoft
       ^^^^^^^^^^^^ ^^^^^^

I know I should get around to updating my Linux version one of these days but
I am holding out until the TCP/IP & NFS get a bit more stable.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
   Mark Spychalla (spy@castlab.engr.wisc.edu)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%