Subject: Re: smp for ss10
To: David Maxwell <david@vex.net>
From: Lubomir Sedlacik <salo@Xtrmntr.org>
List: port-sparc
Date: 02/09/2003 17:42:33
--UBnjLfzoMQYIXCvq
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

hi,

On Sun, Feb 09, 2003 at 11:20:42AM -0500, David Maxwell wrote:
>=20
> Depends which 40Mhz CPUs.
>=20
> http://lios.apana.org.au/~cdewick/data/suncpum.html "1.- The SM-30,
> SM-40 and SM-50 CPU modules have no integral SuperCache, and they are
> not recommended for use in any multiprocessor systems."

that's not the problem.  even some of SM40 modules can do SMP and the
lack of cache is not the show stopper (it will be less efficient, of
course).  the real problem lies in MBus protocol levels.  earlier MBus
modules have support only for the Level 1 MBus protocol which doesn't
define SMP support at all.

from the MBus protocol specs:

"The complete MBus Specification has two levels of compliance, Level 1
and Level 2.  Level 1 includes the basic MBus signals and transactions
needed to design a complete uniprocessor system.  Level 2 introduces
additional signals and transactions needed to design a cache coherent,
shared-memory multiprocessor."

according to Sun System Handbook, at least 501-2219 and 501-2219-04 are
not SMP capable.
(http://sunsolve.sun.com/handbook_pub/Systems/SS10/SuperSPARC_SM40.html)

> If they're SM-41s, I believe the answer is Yes.

yes, SM41s should work fine.


regards,

--=20
-- Lubomir Sedlacik <salo@Xtrmntr.org>                   --
--                  <salo@silcnet.org>                   --

--UBnjLfzoMQYIXCvq
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (NetBSD)

iD8DBQE+RoT5iwjDDlS8cmMRAoT7AJ9JgM/8Cr/y/E1pEypih4JsC0H9pQCbBsD+
32p1tqWN2RgBOdz9xHWm/go=
=++yL
-----END PGP SIGNATURE-----

--UBnjLfzoMQYIXCvq--