I wonder why we haven't heard any yet. Perhaps the "bug" is simply not a problem - perhaps it's just noise.
This is simply nonsense. It's not that hard to test out, and many people have.
Yes, it's a real bug in the chip.
No, you can't work around it in software.
I say this is a more serious bug than the fp bug of several years ago because the number of users likely to actually encounter that earlier bug was very small: the error was a small roundoff error on a tiny number of floating point operations. Many users don't use FP much at all. Yes, if I had been designing airplane wings with the processor I would have wanted a fix. The rest of us were more likely to be struck by lightning than to see any effect of that bug. Yet the bug, and INTC's mishandling of it, knocked something like 30% off INTC's market cap.
The current bug affects anyone running a multiuser server (Windows NT, FreeBSD, some UNIX-like OS starting with an "L", Solaris, ...). Any unprivileged user on such a machine can crash the whole machine. And there's no way to stop them.
Yes, of course Intel can survive this. The Pentium is nearing the end of its lifecycle, anyway, and Intel can certainly afford to fix the problem and replace a few tens (or hundreds?) of thousands of CPU chips. But much will depend on how adeptly Intel handles the spin. And it's bound to give a little bit of a boost to AMD and Cyrix, whose chips are faster, cheaper, and don't have this bug.
Personally, I sold all my INTC Friday at 77, as soon as I'd verified the bug. I'll be looking to step back in somewhere below 75. |
