Re: which AMD chips are bad - if anyone cares <g>

Juhani Rautiainen (jrauti@sasu1.carelian.fi)
Tue, 9 Sep 1997 18:19:35 +0300 (EET DST)


On 8 Sep 1997, Benoit Poulot-Cazajous wrote:

>
> R.E.Wolff@BitWizard.nl (Rogier Wolff) wrote :
>
> > Date: Sun, 7 Sep 1997 15:47:36 +0200 (MET DST)
> >
> > Somebody pointed me towards:
> > http://www.amd.com/K6/k6docs/revgd.html
> >
>
> On my K6-200 (B 9720...) "certain another relative internal pipeline timing
> conditions" are not satisfied.
>
> > /* K6 FP bug detection routine by R.E.Wolff@BitWizard.nl */
> > int testk6 (void)
> > {
> > double b = 0;
> > int i;
> >
> > for (i=0;i<6;i++)
> > b = (b + 0xd34) * 4096;
> >
> > return (b * 5) == 0;
> > }
>
> My K6 does not have this bug, even if it has the >32 MB problem...
>
>
> How many kinds of K6 with the same stepping are there ???
>

Did you compile with -O2 flag. Without -O2 it gives you OK-message. I've
got B9730 series chip which seems to have this bug so it should be in
9720 series if it isn't introduced later.
I'll try your cpuid program tonight but last night I checked those
cpuid compinations and there isn't many places which could be different
if you check AMD documentation. I think there is some hidden method
in finding out different processor revisions under same stepping.

Juhani

--
Juhani Rautiainen			jrauti@iki.fi