Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.arch > #6246

Re: IEEE-754-2008 2's complement question

Show all headers | View raw

John Levine wrote:
> 
> How do you know it's a denorm?  In binary floating point, you know
> because it has a zero exponent.  This hack, as I understand it, has
> .1000-.1999 (norm) .2000-.9999 (denorm), for any exponent.
> 
> If that's not what you mean, you're not explaining very clearly.

No, that is a fatal defect to the scheme.  When I
first considered the problem, I briefly thought that
value-dependent variable-precision is fatal, but
actually it's not, or it shouldn't be.

What Benford's law exposes is that the amount of
information encoded in a number representation is
not a constant for a given number of digits (binary
or otherwise), nor is it monotonic.

I've tried to think of a practical representation
that would be monotonic, but I haven't come up
with anything yet.  If there were such an encoding,
it would come closer to the limit on how much
information can be stored in a given number of bits.

Back to comp.arch | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

IEEE-754-2008 2's complement question Daku <dakupoto@gmail.com> - 2012-03-04 05:12 -0800
  Re: IEEE-754-2008 2's complement question Mark Thorson <nospam@sonic.net> - 2012-03-04 10:37 -0800
    Re: IEEE-754-2008 2's complement question BGB <cr88192@hotmail.com> - 2012-03-04 13:21 -0700
      Re: IEEE-754-2008 2's complement question BGB <cr88192@hotmail.com> - 2012-03-04 13:49 -0700
    Re: IEEE-754-2008 2's complement question Quadibloc <jsavard@ecn.ab.ca> - 2012-03-04 13:08 -0800
      Re: IEEE-754-2008 2's complement question BGB <cr88192@hotmail.com> - 2012-03-04 14:43 -0700
      Re: IEEE-754-2008 2's complement question Mark Thorson <nospam@sonic.net> - 2012-03-04 16:35 -0800
        Re: IEEE-754-2008 2's complement question John Levine <johnl@iecc.com> - 2012-03-05 01:39 +0000
          Re: IEEE-754-2008 2's complement question Mark Thorson <nospam@sonic.net> - 2012-03-04 18:53 -0800
            Re: IEEE-754-2008 2's complement question John Levine <johnl@iecc.com> - 2012-03-05 04:04 +0000
              Re: IEEE-754-2008 2's complement question Mark Thorson <nospam@sonic.net> - 2012-03-04 21:13 -0800
                Re: IEEE-754-2008 2's complement question Terje Mathisen <"terje.mathisen at tmsw.no"> - 2012-03-05 08:27 +0100
                Re: IEEE-754-2008 2's complement question BGB <cr88192@hotmail.com> - 2012-03-05 01:21 -0700
                Re: IEEE-754-2008 2's complement question Jean-Marc Bourguet <jm@bourguet.org> - 2012-03-08 16:57 +0100
                Re: IEEE-754-2008 2's complement question Terje Mathisen <"terje.mathisen at tmsw.no"> - 2012-03-08 18:55 +0100
                Re: IEEE-754-2008 2's complement question Stefan Monnier <monnier@iro.umontreal.ca> - 2012-03-05 17:01 -0500
        Re: IEEE-754-2008 2's complement question Joe Pfeiffer <pfeiffer@cs.nmsu.edu> - 2012-03-04 22:32 -0700
  Re: IEEE-754-2008 2's complement question MitchAlsup <MitchAlsup@aol.com> - 2012-03-04 11:46 -0800
  Re: IEEE-754-2008 2's complement question bert <bert.hutchings@btinternet.com> - 2012-03-04 12:05 -0800
    Re: IEEE-754-2008 2's complement question dakupoto@gmail.com - 2012-03-04 19:40 -0800
      Re: IEEE-754-2008 2's complement question Joe Pfeiffer <pfeiffer@cs.nmsu.edu> - 2012-03-04 22:33 -0700
      Re: IEEE-754-2008 2's complement question bert <bert.hutchings@btinternet.com> - 2012-03-05 02:16 -0800
  Re: IEEE-754-2008 2's complement question Quadibloc <jsavard@ecn.ab.ca> - 2012-03-04 13:09 -0800

csiph-web