> > I can see all sorts of problems with such a scheme, but if the machine
> > ever worked, someone clearly found solutions....

I've heard a number of different stories over the year about how this
worked.. 

(the one i believe was that normally one processor was running, and
the other was sitting idle; when a page fault occurred, the first
processor was "frozen", and the second processor ran the fault
handler, paged in the missing page, and then unfroze the first
processor.. the first processor just thought it had been talking to a
memory location with tens of thousands of wait states..)

> Yes, old Apollo computers were built with a similar architecture.  It
> was the DN3XX's I think.  

DN3xx's were single-piece desktop units with the monitor and cpu in
the same box.

The DN1xx's and DN4xx's were the small-refrigerator-sized units.

					- Bill