Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!franco
From: franco@iuvax.cs.indiana.edu
Newsgroups: comp.sys.atari.st
Subject: Floats in GCC v1.3x
Message-ID: <36500063@iuvax>
Date: 8 Mar 89 00:42:00 GMT
Organization: Indiana University CSCI, Bloomington
Lines: 22
Nf-ID: #N:iuvax:36500063:000:863
Nf-From: iuvax.cs.indiana.edu!franco    Mar  7 19:42:00 1989


Here is a question concerning floating point capability using gnu c-compiler
v 1.34.  I cannot seem to be able to get floats having value greater than
10^9 or less than 10^-9.  For example, the program below gives erroneous 
results for z and c:

main()
{  double x,y,z,a,b,c;

   x=1e5; y=1e8; z=1e12; a=1e-5; b=1e-8; c=1e-12;
   printf("\n\n%le %le %le %le %le %le %le\n\n",x,y,z,a,b,c);
} 

In fact, the output is not even in e format (but f format instead).  The 
question is, am I doing something wrong (possibly an incorrect switch when
compiling) or is there essentially no float support with the gcc so far
ported to the atari?  If floats are currently broken, is there a way that
I might be able to fix them?

BTW I have tried several switch combos including ffloat-store, -DIEEE etc.

Email to franco@iuvax.cs.indiana.edu would be appreciated.  Thanks.