This is the full text published in the 6. December 1994
edition of the "New York Times":



ERROR IN PENTIUM CHIP STILL DOGGING INTEL

By JOHN MARKOFF


(c) 1994 by N.Y. Times News Services, USA.



   BURLINGAME, Calif. -- A math error found in all of Intel Corp.'s
top-of-the-line Pentium microprocessors is proving to be a continuing
public relations nightmare for the world's largest chip maker.

At a computer industry conference that began in Burlingame on Monday,
two of the largest computer makers attempted to minimize the growing
public alarm about the error the chip makes in some rare cases while
doing high-precision division operations.

But at the same time, Compaq Computer Corp.'s chief executive, Eckhard
Pfeiffer, said his company planned to make available a software fix
that would turn off the faulty hardware component of the chip and give
the user a slower -- but accurate -- solution until Intel begins
supplying new chips without the error early next year.

"There's no panic," Pfeiffer said. "I don't think there is any
reason for a general product recall, but if the customer is worried,
this is something to fix immediately."

The IBM executive in charge of personal computers, G. Richard Thoman,
said that while for most people there was no issue from the error, his
company was still attempting to determine the scope of the problem.

"I don't want to be alarmist, but for certain kinds of calculations
we're not sure what the boundaries of the problem are," said Thoman,
a senior vice president and group executive at IBM.

Last week IBM said that it would work with customers who wished
replacement parts.

Both executives said their companies had not been receiving large
volumes of calls. Separately, the chief executive of Gateway 2000, Ted
Waitt, told an audience at an investment conference in San Francisco
that his company's shipment rate had shown no signs of letting up in
the face of growing concern about the flaw.

However, a survey published Monday in Computer World, a weekly
computer industry newspaper gave a very different view of the concerns
of corporate computer managers.

A survey of 100 corporate managers found that 65 percent were either
very or somewhat concerned and that 68 percent said they felt that
Intel's case-by-case replacement policy was unfair. Moreover, almost
half of those responding said they intended to delay purchases until
the problem was resolved.

Intel officials discovered the problem this July but decided it was
not significant enough to announce publicly. The error became widely
known last month after Thomas Nicely, a math professor at Lynchburg
College in Lynchburg, Va., sent a private electronic mail message to
several colleagues asking them to check their machines for the error.

Industry analysts said that all six million Pentium-based personal
computers expected to be shipped this year will have the flaw.

Industry executives and technical experts said they were less
concerned about the actual hardware error than by the way Intel was
responding.

"Intel is to be faulted for their lack of disclosure rather from the
fault itself," said W. Jerry Saunders 3rd, the chairman of Advanced
Micro Devices Corp., Intel's chief microprocessor competitor.

"They seem to be in denial mode, which is crazy," said Michael
Slater, editor of Microprocessor Report, a computer industry
newsletter. "They took the gamble that they could wait until
corrected silicon was available, but they lost."

Intel said it would worry about how it had handled the affair after it
had finished dealing with the immediate consequences of the uproar.

"A few weeks from now we'll reflect and see what we goofed up this
time around," said Howard High, an Intel spokesman. "Now we're
concentrating on mobilizing the company and making sure all the people
that need to be responded to get handled quickly."

Another computer company that had a similar problem many years ago
responded very differently, according to Richard Doherty, publisher of
Envisioneering, a computer industry technical newsletter.

He said that when he was a college student in 1973, Hewlett-Packard
had found a multiplication error in its HP-35 calculator that affected
the accuracy of calculations beyond five significant digits. The
company bent over backward to reach its users to tell them of the
error and offer to replace the machine.

In trading Monday, Intel stock gained $1.25, to $64.125, on the New
York Stock Exchange.

-------------------------------------------------------------------------

