Computer History: Intel Pentium Flaw

The computer industry went through many difficulties along its long path towards what we see in today’s innovations, this also includes processor industry leader, Intel.

In this post I try to cover one of the major technical challenges that faced Intel during the launch of its new Intel Pentium processor, what did the CEO do? And how did Intel eventually handle it?


History

The story begins in the year 1993, when Intel released the first generation of Pentium processor as a successor to the i486, bringing significant performance enhancements. Unfortunately, that version had a design flaw that caused social and ethical issues, rather than technical ones. Most researchers exaggerate and refer to it as the Intel crisis; it was called so not because of the flaw itself; because technical errors or bugs happen all the time in IT, but because of how it was handled or mishandled.


Discovery

In 1994, Thomas R. Nicely, a professor of mathematics at Lynchburg College, discovered a bug in the floating-point arithmetic unit that could render wrong answers when dividing numbers. As a researcher, he worked hard to troubleshoot it, he changed PCs, tried different processors, motherboards, and programming languages, eliminating all other factors until he had no doubt, it was the Intel Pentium processor. He documented his findings and tests of the bug and sent them to Intel technical support who promised an answer in a few days but didn’t.


Media

To confirm if others had the same problem, he shared the issue with a couple of industry professionals. The bug, then, was circulated online in specialized forums; later, it was published in magazines, very soon it was picked up by CNN and went public in the news. As usual, the media could not describe it accurately to their audience, and the conveyed message was that processor giant Intel had a design flaw in a CPU design and they are refusing to admit it as if Intel was dishonest to their customer.

Solutions

After inspecting the bug, Intel confirmed it and published a 32 pages whitepaper describing its causes in detail and emphasizing the fact that very few users and applications would be affected. The whitepaper statistics and calculations showed that the average PC user is likely to encounter a failure once every 27,000 years due to this flaw.

Intel’s policy at that time was that the end-user should be able to demonstrate the error to get a replacement. Usually, such an offer should be okay solving the problem, for Intel, it is the first time for them to deal directly with clients. According to Andrew Grove, Intel’s CEO, that policy was almost working, and complaints were dropping over time.


Competition

On the contrary, IBM claimed that their tests showed that spreadsheet users might encounter this issue every 24 days. However, Professor Nicely eventually agreed with Intel’s findings confirming that the error is not that common as IBM described it. Nicely also stated that the error is not of consequence to most users, arguing that after over a year of worldwide use of Pentium systems, not a single one of roughly a million users had noticed it.

IBM logo

In December 1994, IBM stopped selling Intel Pentium PCs until Intel declared defeat and offered all users a full replacement of any Intel Pentium CPU. Although Intel and IBM are very close partners, and IBM has adopted Intel’s technologies since 1981, it was not clear why IBM took that decision. Some say they were trying to give their processor a little boost, others claimed that other PC manufacturers had gained a bigger market share than IBM’s, such a decision would have adverse effects on the PC sales in general.


What should have been done?

Users, journalists, and researchers to this day blame Intel for its first reactions to the incident; reports say that Intel was already aware of the issue since June 1994, as it was discovered during the testing process of the upcoming CPU generation. They made a mistake not sharing the issue details when they found it; they should have shared it with the public or at least with their OEM partners; again that is because Intel considered it very minor. Computer manufacturers are the ones who deal with end user’s issues and provide the first level of support, and they design their solutions based on Intel’s recommendations and guidelines.

Intel CEO: Andy Grove
Intel CEO: Andy Grove

The brave decision

Andrew, Intel’s CEO at that time, took a brave decision by recalling and replacing all affected CPUs only after six weeks from the time the issue went to the public, that decision cost the company $475 million but was the reason to what made today’s Intel.


What do you think of this story? Would it be handled differently if the media was not involved? How long would a similar issue take to resolved nowadays with the presence on the internet and social media? Did Intel take the right decision? Let me know your thoughts.