Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!ames!trident.arc.nasa.gov!yee
From: yee@trident.arc.nasa.gov (Peter E. Yee)
Newsgroups: gnu.gcc.bug
Subject: Floating point problems -- ISI68K gcc-1.35
Message-ID: <26864@ames.arc.nasa.gov>
Date: 12 Jun 89 15:54:55 GMT
Sender: usenet@ames.arc.nasa.gov
Reply-To: yee@trident.arc.nasa.gov (Peter E. Yee)
Distribution: gnu
Organization: NASA Ames Research Center, Moffett Field, CA
Lines: 36

I found the following differences in my stage1 and stage2 output for file
insn-output.c.  Most of ther numbers seem to related to pi and sqrt(2).
Why would I get different values in the last few decimal places.  This is
on an Integrated Solutions 68030 system, running ISI 4.3 BSDUNIX version 5.01.
We have the Green Hills C compiler installed.  Any help appreciated.

						-Peter Yee
						yee@ames.arc.nasa.gov
						ames!yee
PS Compiling with -msoft-float eliminates this problem, but it's not the
solution I'm looking for.

1213c1213
< 	fcmpd #0r6.2831854820251473726,fp0
---
> 	fcmpd #0r6.2831854820251464844,fp0
1217c1217
< 	fcmpd #0r3.1415927410125736863,fp0
---
> 	fcmpd #0r3.1415927410125732422,fp0
1221c1221
< 	fcmpd #0r1.5707963705062868875,fp0
---
> 	fcmpd #0r1.5707963705062866211,fp0
1233c1233
< 	fcmpd #0r-1.5707963705062868875,fp0
---
> 	fcmpd #0r-1.5707963705062866211,fp0
1257c1257
< 	fcmpd #0r0.4342944920063019687,fp0
---
> 	fcmpd #0r0.43429449200630187988,fp0
1296c1296
< 	fcmpd #0r1.4142135623730951899,fp0
---
> 	fcmpd #0r1.4142135623730951011,fp0