Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Bill Jackson who wrote (84318)	12/29/1999 7:59:00 PM
From: Elmer	Respond to of 1572973

Re: "With enough care bugs can be eliminated. Remember there are just so many commands. " Yes there are only so many commands but there are nearly an infinite number of combinations of commands, data sets, modes, interrupts, flags, exceptions, snoops, snarfs, stalls, illegal instructions and on and on... plus all the permutations of the above. It is naive to think it is simple and easy to validate. Ever processor has bugs. Some companies report them to the public and some companies are too ashamed to. EP

To: Bill Jackson who wrote (84318)	12/29/1999 11:28:00 PM
From: Petz	Read Replies (1) \| Respond to of 1572973

Bill, re:<Would not errata show up in the talk areas and comments to OEMs that might leak> Only if they are moderately serious. The K6-2 never had any errata that showed up on discussion boards or USENET. The K6 did. There was a problem that showed up when you had 64M of RAM or more and did a lot of multitasking, such as running 10 compiles of the Unix kernel at the same time. That bug had to be fixed via a mask revision. <Or can AMD stay on top of it by constant BIOS and microcode tweakery?> As I pointed out with Intel, the chances of being able to fix a bug via BIOS are slim. AFAIK, the Athlon doesn't have a writeable control store. Despite the fact that Pentium II and following had some limited writeable control store, I've never seen a bug listed in the document being solved by using that feature. Neither Intel nor AMD writes the BIOS except maybe the first one. The way to get all the mobo manufacturers to implement a BIOS fix is to 1)document the problem in the Specification Revision (Intel lingo) or the "Revision Guide" (AMD lingo), then 2)write sample code which solves the problem. In Intel's case, since they make motherboards for public purchase, when they solve a processor errata using a BIOS fix (rare, as I said), they document it in the Specification Revision for the motherboard. <Has anyone compared BIOS from the majors, over time to see if they have changed to hide a bug?> I haven't checked to see if Asus, Biostar, Abit and fifteen gazillion others have bothered to implement the fixes for processor bugs in their motherboards. I suspect that Coppermine bugs E44 and E46, which can cause system hangs, are only fixed in Intel motherboards at this point. Petz