If you go to Intel's developer website, they have a series of documents
covering the SMP spec and steppings for their CPU's, referred to as
Processor Specification/Updates. I remember running 2 90 and 2 100 MHz
parts. I avoided the B steps. Now i'm running 2 200MMXes on a TomcatIVd.
You have to read through the errata carefully, especially if you are doing
kernel level work. Certain older steppings, even when matched, have
problems with pipeline stalls, misses, next address assertions, etc.
http://developer.intel.com/design/pentium/specupdt/
Grab the appropriate Specification Update PDF file.
Robin Chung [EMAIL PROTECTED]
http://www.motogp.com [EMAIL PROTECTED]
ph: +44 (0)7971 575 394 [EMAIL PROTECTED]
fax: +44 (0)7970 573 175 Scully to Mulder: "I had you big time..."
On Wed, 7 Jul 1999, Alan Cox wrote:
> > If I compare differences between those files, that are supposed
> > to be the same, I obtain :
>
> > 5649c5649
> > < 00169b0 4697 9c90 6e6a ccbf 77ea bd59 cc2b 6186
> > ---
> > > 00169b0 4697 9c90 ee6a ccbf 77ea bd59 cc2b 6186
>
> 6e->ee is 1 bit (01101110 -> 11101110)
>
> > 21770c21770
> > < 0055940 5a9d ad36 37a4 bb45 3144 c56c 4c70 e7a4
>
> 4c->cc is 1 bit (01001100 -> 11001100)
>
> > > 0055940 5a9d ad36 37a4 bb45 3144 c56c cc70 e7a4
>
> > WARNING: SMP operation may be unreliable with B stepping processors.
> > enabling symmetric IO mode... ...done.
>
> B step Pentiums. Thats ominous anyway.
>
> > I guess that B stepping processors could be the cause of the problem.
> > I read the doc in ./arch/i386/kernel/smp.c but this is not very clear
> > to me. Does a workaround exist for this case ? Thanks for your help.
>
> The B step pentiums bugs wouldnt explain this. You'd expect to lose IRQs,
> or see the machine freeze. One of those CPU's is very old though.
>
> With the Pentium the rules for pairing chips are very explicit. The early
> B series chips also have problems with REP MOVS on 100Mhz parts and the like.
> The errata document makes interesting reading.
>
> I don't believe B1 and B6 is a legal pairing of CPU's. Intel should be able
> to tell you if it is. The rules for Pentiums especially old ares quite limiting
> compared to the current CPUs.
>
> Alan
>
>
> -
> Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
> To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
>
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
