Jim Nelson
10 May 1997

You will be assimiliated.
Resistance is fu ... fu-till ... five?

Never fear, though. Intel promised to ease worries with their fact-finding report from their own internal blue ribbon committee. I was confident that Intel's marketing staff, er, "committee" would soak the report in enough syrup and sugar to make it palatable for even the most conservative Chief Information Officer. Indeed, they sugar-marinated it with a liberal combination of technical jargon, professional concern, and third-party statements of support.

But why listen to Intel? Robert Collin's has laid out the whole sordid affair, right down to demonstrating which opcodes are faulty. If you're not of the technical persuasion, just skip down to the bottom of the page, where Robert summarizes: "Since I'm not a numerical analyst, you should draw your own conclusions ... [however], the Pentium Pro and Pentium II processors behave differently than their predecessors."

The last time Intel flubbed their floating-point microcode, their well-oiled marketing machine fell apart like a cheap wristwatch. Instead of fessing up, Intel stonewalled. When it became really damn obvious the Pentium's FDIV instruction was buggy, they admitted the issue and then ... stonewalled some more. "People shouldn't worry. The problem only arises when doing really intricate math operations."

Intel's not going to try that route again -- but from their "report", it's obvious they have no intention of recalling the chip or replacing anyone's CPU in the future. This'll get a little blurb on their errata sheet and that's it. After all, people using Excel or playing Interstate '76 won't notice a thing. (People aboard a jumbo jet designed with Pentium II workstations, nose-diving into the Pacific, might have a valid concern however.)

Some Intel wonk said to the press "no microprocessor is perfect." Yeah? Well, I've got a suggestion for Intel. Take a hard look at your marketing budget. Take a hard look at the money and resources burnt to design the way-new (and way-proprietary) Single Edge Contact packaging. Take a hard look at the lawyer's fees to ward off the n + 1 class action suits stemming from the old Pentium bug. Compare them to your Quality Assurance budget. Take your time, I know this is a toughie.

Is it just me, or is Intel rapidly adopting all the ugly behaviors of American car companies in the Sixties? Big, defiant, and seemingly unstoppable. Redesigning models every other year to get Mom and Dad back on the showroom floor. Dumping huge amounts of cash into whiz-bang product launches.

Which makes the Pentium FDIV problem Intel's Corvair. So what's this new one? Intel's version of that tailgatin' nightmare, the Ford Pinto?