[rescue] Re: SM[567]1/HyperSPARC comparison

[rescue at sunhelp.org]

The MBus guide does have the straight gouge on
this.  As Dave said, mixing speeds works--and since
each MBus module has an onboard oscillator, they
do each run at their own speed.  The only other
considerations are the MBus speed (20s can go faster
than 10s, even with the same MBus modules, IIRC) and
the cache controller type.  I believe that I
tried one of my SM71 pair in the same system with
my SM61 and had no joy--because of different cache
controller revisions.  Even within the SM71 "family"
there are modules that can't live together in peace
within the same system.

If we do get to the point of trying to do a complete
set of benchmarks, I've got a grunch of systems to
throw into the pot from SS2 through E3000 and SB100.
Let's seriously try to find a good benchmark suite
that is both freely available and useful for real
comparisons.

  --Rip

-----Original Message-----
From: Kurt Mosiejczuk
To: rescue at sunhelp.org
Sent: 2/5/2002 2:49 PM
Subject: Re: [rescue] Re: SM[567]1/HyperSPARC comparison

On Tue, 5 Feb 2002, Dave McGuire wrote:

>   I've mixed speeds in mbus machines plenty of times.  It works fine.

I think http://mbus.sunhelp.org/ has the best sounding advice on this.

"You should not mix MXCC cache-controllers of differing major revisions,
on the same MBus/XBus."