On Wed, 29 Dec 1999, Roger Ivie wrote:
> Platlyn said:

> > Could NetBSD be made to use =
> > them? 
> 
> Probably not. Such a system would not be a symmetric multiprocessor;
> the additional KA630s would each need their own memory (which means
> you'd need a custom backplane to build the system in; basically, each
> KA630 would have to have its own set of CD slots for the PMI).
> Fundamentally, the multiprocessor support in the KA630 consists of a
> way to use an external bus arbiter and a doorbell register.

It might be possible to use a second KA630 as a coprocessor of sorts.  
I'm sure we've had a discussion like this on the list within the past
couple of years.  You could run the second KA630 using the 1Meg of memory
it has on-board.  It definitely wouldn't be an SMP arrangement.

In the case of the BA123 would it be possible to split the CD backplane
slots 1-4 into two groups of slots (1-2 and 3-4); then have one
KA630+Memory in slots 1-2 and a second KA630+Memory in slots 3-4?  Though
I think that would require some modification to the backplane. :-)  Or
what about chaining two complete MicroVAX II BA23 systems together using
the M9404 and M9405-YB Q-Bus interconnect modules?

And then I wonder how difficult it would be to implement IP over Q-Bus?
You could run multiprocessing tasks using PVM or MPI.

-brian.
--- Brian Chase | bdc@world.std.com | http://world.std.com/~bdc/ -----
      Are you absolutely sure that you want to do this? [ny] m