Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!oliveb!ames!lll-winken!vette!brooks
From: brooks@vette.llnl.gov (Eugene Brooks)
Newsgroups: comp.arch
Subject: Re: Question about IEEE floats
Message-ID: <27380@lll-winken.LLNL.GOV>
Date: 23 Jun 89 05:53:17 GMT
References: <67700005@sts>
Sender: usenet@lll-winken.LLNL.GOV
Reply-To: brooks@maddog.llnl.gov (Eugene Brooks)
Organization: Lawrence Livermore National Laboratory
Lines: 17

In article <67700005@sts> roy@sts.sts.UUCP writes:
>
>I would like to find out about the IEEE floating point format and,
>specifically, about the "hidden bit".  I do know that, for example,

The situation is very simple.  All normalized numbers
either have a 1 bit as the most significant bit or are ZERO.  ZERO is the
special 32 or 64 bit word of all zero bits, and all other normalized
numbers have in the most significant bit slot a 1 so we don't bother
chewing up a bit of memory for it.  We get one more significant bit for
the mantessa for our trouble.  This bit which is not present in the memory
representation is called the hidden bit, and it gets stuck back in and taken
back out as things go in and out of the functional units.  For a denormalized
floating point number, the hidden bit is, you guessed it, a 0.


brooks@maddog.llnl.gov, brooks@maddog.uucp