On Mon, Feb 02, 1998 at 08:18:49AM -0800, Michael L. VanLoon -- HeadCandy.com wrote:
> 
> >Identifying Cyrix chips can be a bit confusing. I think the issue here
> >is that the 6x86 is almost, but not completely a 586. Also, it doesn't
> >have cpuid enabled by default (which makes it incompatible with the
> >Pentium as well), but does have the feature.
> 
> This is kinda silly.  Saying the 6x86 is not quite a 586 because it
> doesn't properly implement a cpuid instruction (which if I'm not
> mistaken, the first Pentiums didn't have, either), is absurd.  The
> 6x86 is clearly a more advanced design than the Pentium.  To be
> correct, technologically, the 6x86 is more like a 5-1/2x86.  Obviously
> it's not a 686, but it's more than a 586.

First of all, I did not say that the 6x86 is not considered a 586 because
it does not have cpuid enabled by default. There are other differences.
The "class" identifier is used in the kernel to say "it is 100%
compatible with at least that class CPU from Intel". Now, I have
nothing against the Cyrix chips, but saying "clearly more advanced
design" is not a clear flag to base decisions in the kernel on.. You
want to know _exactly_ what it matches with. And since this all
started with Intel chips, that's the standard that is matched against.

> True, but it's disconcerting for a user, who may not know anything
> about these distinctions, to have his nice "Pentium-like" box called a
> 486.

If it's confusing to users, perhaps I should just remove the XX-class
message, it isn't of much use anyway.

- Frank