Dear DDJ,
Have you heard about the fdiv bug in the Intel Pentium chip that Thomas R. Nicely, a mathematics professor at Lynchburg College, ran across? The problem is that the Pentium produces faulty numbers in the ninth place to the right of the decimal point and beyond. Nicely found it when checking double, triple, and quadruple prime numbers--specifically in analyzing properties of twin prime numbers. Although Intel says the average user should only run into this problem once every 27,000 years, they're still offering to replace buggy chips if you call 800-628-8686 and demand new ones. Example 1(a) is a Basic and Example 1(b) a C program that test whether or not a computer is using a chip with this bug.
Harry J. Smith
Saratoga, California
DDJ responds: Thanks Harry. Yes, we heard about the fdiv problem. We also heard why Intel didn't name the "Pentium" the "586." (When they added 486+100 on the first Pentium, they got 585.999983605.)
Dear DDJ,
Although well versed in a number of computer languages, I'm not up on what appears to be the Latin in Jonathan Erickson's January 1995 editorial--"Cursor sine termino." What the heck is that all about?
Zac Davis
Austin, Texas
DDJ responds: Yep, it's Latin, and it literally means "a runner with no limits," the formal way of saying "running light without overbyte." Thanks to our staff Latin expert (the publisher's mother).
Dear DDJ,
In the article, "Building an E-Mail Manager" (DDJ, December 1994), Michael Floyd praised Qmodem from Mustang Software, as do I and many people I know. However, readers should note that Mustang Software Inc. is known for its Wildcat! BBS package, not its Mustang BBS, as the article states.
As a Wildcat! BBS System Operator, I wanted to point this out. Please keep the quality articles coming. Maybe even do a story about the new programming language that is included with the latest versions of Wildcat! (4.0 and 4.01).
Dave Noice
Columbus, Ohio
DDJ responds: Thanks Dave. We thought fellow readers might like to know more about your BBS. Super-Port is a BBS Service Bureau, providing BBS services to individuals, associations, and businesses. The only BBS Super-Port runs for itself is for customer demonstrations (Wildcat!). To arrange for demonstrations, or to request information packets, call Super-Port at 614-385-2003. You can also e-mail requests to: dave.noice@commport.org; include your name, address, phone numbers (voice and FAX) and e-mail address.
Dave uses Wildcat! BBS software for most of the system, but has the ability and expertise to use any commercial BBS package currently available. He provides complete BBS packages (turnkey systems), including hardware, software, custom display screens, menu design, file and message areas, and conferences.
Dear DDJ,
With regards to Homer Tilton's letter, "Mind and Life" (DDJ, December 1994), I'd like to take issue with his ad hominem attack on those who feel that mechanistic models of human thought are inadequate. One could argue just as convincingly that proponents of such models find them comforting because they can understand them without having to think too hard.
In any case, both he and Michael Swaine seem to have missed part of the point of Rodger Penrose's The Emperor's New Mind (and the recently published Shadows of the Mind). Penrose is not arguing from quantum uncertainty (being one of Britain's leading mathematical physicists, I suspect that he is as familiar with the literature as Mr. Tilton), but rather from the noncomputability of the deterministic quantum laws. He is not a mystic (as he takes some pains to explain in Shadows), but simply believes that there are aspects of human awareness and understanding (such as the proofs of Gödel's Incompleteness Theorems) that transcend what a Turing Machine can accomplish and that such transcendence is mediated by a physical phenomenon whose workings further research can bring to light.
There is great comfort in thinking that we know everything about a particular area of knowledge, but we should recall the feeling prevalent in the late 19th century that "everything has been discovered and it only remains to work the laws out to a few more decimal places." History teaches us that shortly afterwards, Planck, Einstein, and Dirac revolutionized physics, while Gödel, Turing, and Church demolished David Hilbert's ambitious program to mechanize all of mathematical reasoning. Penrose may not be right (as he freely admits) but he is at least open to the possibility that we do not yet have all the answers--or even all the questions. I personally find this attitude to be more intellectually stimulating than wandering around with a mechanistic hammer treating the world as the unfortunate metaphorical nail.
Richard Wesley
Seattle, Washington
Dear DDJ,
In his December 1994 "Swaine's Flames," Michael Swaine asked us to guess what two people were represented by:
&8-) 7 and (:-\ L
I may be wrong, but this looks suspiciously like the occupants of 1600 Pennsylvania. The &8-) 7 is Hillary with a bow in her hair. The (:-\ L is Bill frowning about Newt as the new Speaker of House.
I hereby request my 15 minutes of fame.
Vance Rigg
Newport Beach, California
How about Penn & Teller?
P.S. What does the "CyberSpeak Pronunciation Guide" say about the pronunciation of "emoticon"? Or is it one of those words to be seen and not heard? No, no. I've got it! Siskel and Ebert! I could tell from the bald pate and the glasses_.
Kerry Burton
kerrykjb1@aol.com
My answers for who those emoticons represent are: Somebody bald & somebody with hair.
Kevin Haskel Rubin
gnome@teleport.com
Is it Siskel and Ebert?
Jerry Chadwick
jerry_chadwick@Novell.COM
Yo, Mike:
My guess for the smiley contest: Penn and Teller. (The top smiley is Penn.) By an amazing coincidence, I'm now working on a book of Teller's collected quotations. (I've already finished a similar work on Harpo Marx.)
Mark Gingrich
st190022@s1.csuhayward.edu
*** This letter was assembled from 100% recycled bits. ***
&8-)7 (:-\L
Looks like Laurel and Hardy to me.
Scott Redding
sreddin@apg-9.apg.army.mil
It looks like Siskel and Ebert, thumbs up and thumbs down!
Todd Hale
thale@novell.com
() () () () /[]\ /[]\ /[]\ /[]\ .-##..___.##--..__##.---..##_..-- .-|//___..|//.___.|//-.___|//--. .---.___.---.___.---.___.---.__ .---.__.---.__.---.__.---.__.---
Siskel & Ebert:
&8-) 7 is Siskel (or, the fat one_I forget which one is which)
(:-\ L is Ebert (or, the skinny one_I forget which one is which)
Gregg Cooke
gcooke@rt66.com
(:-\L = Stan Laurel
&8-)7 = Oliver Hardy
Denis Blodgett
blodgett@monet.vill.edu
Siskel and Ebert. But they only get quoted when they agree.
John Maxfield
73523.736@compuserve.com
The emoticons represent Laurel & Hardy. Thank you.
Jay Joiner
70263.1054@compuserve.com
Laurel and Hardy perhaps? :-}
N N E E E E ------------------------------------- N N N E N E E D H A M ' S E L E C T R O N I C S N N N E E E Device Programmers N N N E ------------------------------------- N N E E E E
Eric Cox
needhams@crl.com
&8-)7 + (:-\L == Oliver Hardy + Stan Laurel
tamortir@cris.com
Unknown @ Cogitate Inc.
My guess is &8-) 7 is Bill Gates, and (:-\ L is/are his lawyer(s). Do I win the prize?
Ed Remmell
eremmell@Internet.cnmw.com
Hi Mike,
I have always enjoy your flamorous editorial/commentary article, even though I sometime don't understand all of the fancy terms. On your emoticons. I could guess only one person:
&8-) 7 is Gates,
Who is (:-\ L ? P. Kahn of Borland?
Thanks & how about
O O .^. __, \\ // \//
na...
Lan Tran
LTRAN@CSTP.UMKC.EDU
(a)
10 DEFDBL A-Z: COLOR 14, 1: CLS : PRINT 'Yellow on Blue
20 PRINT "FDIV - This program checks for an error > 10^-15 when using"
30 PRINT "fdiv. When detected it warns you that you probably have the"
40 PRINT "Pentium fdiv bug."
50 PRINT
60 PRINT "GWBASIC Version 1.0, last revised: 1994-11-26"
70 PRINT "Copyright (c) 1994 by author: Harry J. Smith,"
80 PRINT "19628 Via Monte Dr., Saratoga, CA 95070."
90 PRINT
100 FOR i% = -3 TO 3
110 in = 824633702449# + i%
120 ou = (1# / in) * in
130 IF ABS(ou - 1#) <= .000000000000001# THEN GOTO 150
140 PRINT "You have the Pentium bug"
150 PRINT in; "produced an error of"; ABS(ou - 1#)
160 IF ou <> 1# THEN PRINT ou; "<> 1#"
170 NEXT i%: PRINT
180 PRINT "This program is a modification of a C program I found on the"
190 PRINT "Internet in the newsgroup comp.sys.intel, in a message posted by"
200 PRINT "Bill Broadley Broadley@math.ucdavis.edu UCD Math Sys-Admin"
210 'If you get the message "1 <> 1#" without, the message "You have the
220 'Pentium bug" the program may be using floating point emulation and not
230 'checking the CPU fdiv instruction. This is the case for the GWBASIC
240 'interpreter. It is better if you compile this program to FDIV.EXE
250 'before running.
(b)
#include <stdio.h>
#include <math.h>
#define C 824633702449.0
double
test (double x)
{
return ((1.0 / x) * x);
}
void main ()
{
double delta = 1e-15;
volatile double in1, in2, in3, out1, out2, out3;
in1 = C - 1; out1 = test (in1);
in2 = C; out2 = test (in2);
in3 = C + 1; out3 = test (in3);
printf ("Program checks for an error > %e when using fdiv\n", delta);
printf ("When detected it warns you that you probably have the \n");
printf ("Pentium fdiv bug.\n\n");
if (fabs (out1 - 1.0) > delta)
printf ("You have the pentium bug\n");
printf ("%lf produced an error of %e\n\n", in1, fabs (out1 - 1.0));
if (fabs (out2 - 1.0) > delta)
printf ("You have the pentium bug\n");
printf ("%lf produced an error of %e\n\n", in2, fabs (out2 - 1.0));
if (fabs (out3 - 1.0) > delta)
printf ("You have the pentium bug\n");
printf ("%lf produced an error of %e\n", in3, fabs (out3 - 1.0));
}
Copyright © 1995, Dr. Dobb's Journal