Unit 5 Analysis 1 Pentium Flaw

702 Words3 Pages
Pentium Flaw Intel was founded in July 1968 by American engineers Robert Noyce and Gordon Moore. Their Pentium microprocessor was first introduced in 1993. The Pentium was the first Intel chip for PC’s to use parallel processing, which significantly increased its speed. It had 3.1 million transistors on one chip, compared to its predecessor the 80486 with 1.2 million transistors per chip. In 1994 the public began to notice a flaw. Flaws are not uncommon in complicated integrated circuits, and most of them go unnoticed by the user, however, the Pentium flaw was very different. The nature of the flaw was in the floating point math subsection of the Pentium microprocessor, it caused incorrect answers when preforming double-precision arithmetic. In October of 1994 a man named Thomas Nicely, a mathematics professor at Lynchburg collage in Virginia was the first to notice this flaw. He found that the results of some of his calculations were incorrect. He cross checked some of the same calculations on a 486 CPU and the results turned out to be different. Then Tim Coe of Vitesse Semiconductors also discovered some inconsistencies. Testing began to be done on whether or not the microprocessor was flawed. Using spreadsheet software, he was able to take the number, multiply it by another number and then divide it by the same number. If a number is multiplied and then divided by the same number, the result should be the original number. However with the flaw, the result of the calculation was incorrect. I mention this testing due to the fact that it is the most well-known issue of the nature of the Pentium flaw. Professor Nicely sent an email to Intel describing the problem that he had come across. The email stated that there was a flaw in the Pentium Floating Point Unit (FPU). Intel did not respond, so Nicely posted a general notice on the internet asking

More about Unit 5 Analysis 1 Pentium Flaw

Open Document