RISKS note about IEEE-754

[Richard Karpinski <dick@xxxxxxxx>]

Date: Tue, 16 Oct 2007 12:15:12 -0700
From: Henry Baker <hbaker1@xxxxxxxxxxxx>
Subject: Re: Another case of Deploy First, Test Later (Re: Huge,
RISKS-24.85)

"A foolish consistency is the hobgoblin of little minds."  Ralph Waldo
Emerson

Whether intended or not, one consequence of the widespread
implementation of
IEEE-754 floating point arithmetic is that almost every computer now
gets
_exactly_ the same answer, down to the last little bit.  This "answer"
may
be far from the "correct" answer, but at least all of the computers
will be
consistent.  In the "old" (pre-754) days, running the same Fortran
program
on several different computers could uncover potential sources of
error.  No
longer.

(This generic phenomenon of foolish consistency even has a technical
name:
"informational cascade" -- if everyone agrees, then everyone must be
correct.  See http://en.wikipedia.org/wiki/Informational_cascade )

I realize that using multiple types of arithmetic to uncover bugs in
floating point code may not be particularly efficient, but it sometimes
works.  IEEE-754 also provided for rounding modes that would allow for
"range" arithmetic in order to achieve the same ends with much greater
efficiency.  Unfortunately, no one seems to implement or utilize those
rounding modes anymore.

RISKS-LIST: Risks-Forum Digest  Wednesday 17 October 2007  Volume 24 :
Issue 86

ACM FORUM ON RISKS TO THE PUBLIC IN COMPUTERS AND RELATED SYSTEMS
(comp.risks)
Peter G. Neumann, moderator, chmn ACM Committee on Computers and
Public Policy

***** See last item for further information, disclaimers, caveats,
etc. *****
This issue is archived at <http://www.risks.org> as
  <http://catless.ncl.ac.uk/Risks/24.86.html>
The current issue can be found at
  <http://www.csl.sri.com/users/risko/risks.txt>