To ALL: Article...Collins even now admits mistakes...
biz.yahoo.com
Pentium II Bug Report Takes New Technical Twist; Story Posted on EE Times Online at eet.com
MANHASSET, N.Y., May 8 /PRNewswire/ -- The purported floating-point bug involving Intel's (Nasdaq:INTC) Pentium II and Pentium Pro processors has taken an unexpected twist, according to CMP Media's EE Times Online (http://www.eet.com), with a revised technical description coming from Robert Collins, the proprietor of the ``Intel Secrets'' Web site that disclosed the glitch early this week.
Initially, Collins said the bug -- designated ``Dan-0411,'' after the person whom Collins claimed uncovered the glitch and conveyed the information to him -- could occur during overflow conditions involving floating-point addition, when an operand is stored as an integer. An overflow is when a number is too large to be stored inside the chip.
But Collins now conveyed to EE Times Online that the Pentium II and Pentium Pro processors do not provide any indication that an error has occurred.
Collins said he was alerted to this differentiation by Martin Atkinson- Barr, a self-employed physicist in Calabasas, Calif. ``Collins misinterpreted Dan's original (bug report),'' said Atkinson-Barr. ``It's not an error bit -- it's the precision exception bit, which isn't an anomalous condition.''
Atkinson-Barr, who claims to be an expert in numerical-analysis software, said: ``It makes the bug more serious. The way Collins originally described it, the chips were flagging an error -- they were just flagging the wrong error. Now, for a wide range of operands, you won't get an error indication at all.''
At Intel, a spokesman said: ``Until we see what the analysis is, we don't have a comment.'' Intel officials told EE Times that hundreds of engineers are at work investigating the purported bug, in an effort to provide a quick response to the technical community.
Intel has pledged to post a response to the original bug report on its Web site by late Thursday or early Friday.
Many knowledgeable sources maintain that the purported bug is little more than a temporary public-relations black eye. ``I think it (the initial bug report) is much less serious than the 1994 Pentium FDIV bug,'' said Richard Smith, president of Phar Lap Software (Cambridge, Mass.), a vendor of X86 programming tools.
The FDIV bug, which was first reported by EE Times, resulted in a public- relations disaster for Intel, its first-ever chip recall and an eventual charge against company earnings of $475 million.
EE Times Oline is the Web arm of CMP Media Inc.'s EE Times. EE Times delivers news of both business and technology to engineers and technical/corporate managers at electronics and computer systems manufacturers in the United States.
______________________________________________________
hmmmm, the shorts plot thickens.
Michael
|
