Subject: Re: multi CPU uvaxII
To: None <port-vax@NetBSD.ORG>
From: Allison J Parent <allisonp@world.std.com>
List: port-vax
Date: 03/19/1998 14:48:08
<> produce a slave that has more than the 1meg onboard.  A pair of BA23s
<> would work nicely. Box A would have uVAX ram, disks, terminal IO, NI
<> and interconnect to box B with uVAX, RAM and QDSS via q22 cables.
<
<So: if one did this, would both CPUs have access to the drives/DHV/whatha
<would not be sharing memory, we're not talking symmetric.

They can both map to the q22 bus.  They had no ability to share the PMI
interfaced memory.

What I'd proposed was slave server (xterminal, bus interconnected).  
Symetrical processing is limited to the common q22 pool memory (4mb).
Q22 access is 16bits wide and very slow compared to PMI. I don't 
think that would be a performance option.  However a vax as an IO slave
or display client (slave) might have value.  Lots of programming involved.

On the other hand you could track down a 6000 series with 4 cpus and...

<What would be the pinout and max length on a Q22 interconnect cable? I'm 
<plug one end into one Qbus, the other into the other?

M9404/5 with cables, length would be something like 3 (or something) 
meters though shorter the better. 

Allison