LINE: Pentium Joke # 3145727

88 Intel Inside the Cat 888888888888888888888888888888888888888888
88                                                              88
88      See?  My simulation shows the                  _ _ ;\ ._88
88      cat will never jump this high.             .-''    ) `. 88
88      We're perfectly safe up here.            .'_,.  , /     88
88               |              .----,-.-.  \\  / /_ ) /  ` .  '88
88               | __           |::: |  .'  \\\'  \.'         . 88
88 _._         \\-', (`-.___..'>|::: | {     \\. -'        .;.: 88
88') _`-//     @   ,   ` ` ..-' |::: |  `.-.-(        . .'.:.:. 88
88   _   *     //`-._(/   '\    |""" |       //` -._      ` :.; 88
88<)_,-`\\           \   \ \    |::: |      ///.  /_\         ` 88
88                  (| /    \   `----"-.    //  \ `._\ \  \`- . 88
88                   |/   ` \      _    )        `.     `  )    88
88                   \ ' \  ;)    (_3--'           `-.._ _ /   /88
88                _._ `._  .'                             /_.'.-88
88               ;'`-;.._))                              '      88
88               `    ``-'                                      88
88______________________________________________________________88
88// // / // /// // ////// / //// / / / / /// / / /// / // /// /88
88/// / // //  / /// / ///  // / /  // / / / /// // // // / / //88
8888888888888888888888888888888888888 fle...@cse.psu.edu 19941211 88


This picture was an exercise in composition, and didn't quite turn out
the way I wanted.  There were a few things I didn't know how to do, so
I just faked it :).

This started with the phrase "Intel Inside the Cat" running around in
my head, and I tried to work it into a Pentium joke, but it's pretty
much a generic computer-risks joke.
--

hmm.  When I did that cartoon, I thought I was being really late at
jumping on the bandwagon, but then on Monday IBM goes and announces
they're stopping shipments of Pentium PCs.  This is going to get
interesting...

If you're wondering what this is all about, there's an FAQ on the
Pentium FDIV bug at
	ftp://www.isi.edu/pub/carlton/pentium/FAQ
Various Intel and IBM statements are available at
	http://www.intel.com/
	http://www.ibm.com/
And for a history of this mess, see
	http://www.mathworks.com/Pentium/README.html

Quick summary: The Pentium has a flaw in some division operations.
For example, 4195835/3145727 is supposed to be 1.33382045, but the
Pentium returns 1.33373907.  The error is no worse than about 0.006%,
but this can be significant in some cases, especially if the error is
allowed to propagate.

Not all divisions are affected.  Only 1 out of 9 billion pairs of
numbers are "bad", but these bad divisions are clustered in a
non-random way.  Right now, it's kind of up in the air how much of a
problem this is for "normal" users.

Intel says the average failure rate for a typical spreadsheet user is
once in 27000 years; IBM says it's more like once in 24 days.  It's
easy to find ulterior motives for both Intel and IBM's statements, but
the technical issue can be evaluated independent of that.  See the
references above, and the discussions in the comp.sys.intel group.

The main things that people are upset about: Intel discovered the
problem in June this year, but didn't tell anyone about it until it
was independently discovered and publicized in November.  Many
researchers have had to reevaluate the calculations they've been
running on their Pentiums.

And Intel has been sluggish in providing customer satisfaction.
Apparently, fixed Pentium chips won't be available in quantity until
Spring '95.  Intel is offering to replace defective Pentiums sooner
and has given fixed Pentiums to people who have asked, but they want
you to justify your need.  Again, see the references and
comp.sys.intel for details.
--

In article <3cti02$ra...@garuda.csulb.edu> hen...@csulb.edu (Henry Yeh) writes:
>Q: What is another term for the "Intel Inside" label?
>A: Warning label.

	Why did they call it a Pentium, and not a 586?  The used one to add
100 to 486 (486+100) and they got 585.999999967

-- 
=======================================
S. Halliday - sha...@atlas.cs.upei.ca			OS/2   OS/2  WARP

Have you hugged your OS/2 today?