floating point bias

Article View: gnu.gcc.help

Article #323

#323

From: Joe Tien
Date: Fri, 21 Aug 1992 01:39

16 lines
1139 bytes

Hello everyone,

I was looking at the floating point library (floatlib.c) in gcc-2.2.2 and noticed
that the bias (called EXCESS and EXCESSD in gcc's floatlib.c) for single and double precision floating number exponents are 126 and 1022, respectively. The IEEE floating point standard says they should be 127 and 1023. Why is that?

I know this is a stupid question and there must be a very simple answer to it since it can't be a typo (otherwise it would have been caught a long time ago). But I can figure it out and would really appreciate some help on the reason behind
it.

Since this is such a dumb question, you can reply to me at: 
		
		jtien@balboa.eng.uci.edu

Thanks again,

Joe Tien

Technical Details

Message-ID: <2A94493F.10034@orion.oac.uci.edu>

Path: arc1-ca.pugleaf.net!sparky!uunet!stanford.edu!ames!elroy.jpl.nasa.gov!usc!orion.oac.uci.edu!venice.eng.uci.edu!jtien

Show Full Headers

Complete Article Headers

Path: arc1-ca.pugleaf.net!sparky!uunet!stanford.edu!ames!elroy.jpl.nasa.gov!usc!orion.oac.uci.edu!venice.eng.uci.edu!jtien
From: jtien@venice.eng.uci.edu (Joe Tien)
Subject: floating point bias
Nntp-Posting-Host: venice.eng.uci.edu
Message-ID: <2A94493F.10034@orion.oac.uci.edu>
Newsgroups: gnu.gcc.help
Reply-To: jtien@venice.eng.uci.edu (Joe Tien)
Organization: University of California, Irvine
Lines: 16
Date: 21 Aug 92 01:39:11 GMT