🚀 go-pugleaf

>From Lars.Ericson@CMU-10A Sat Feb 20 22:15:51 1982
1) (Random) Has anybody used the new MC68000 machine that came out
(somebody in California, begins with a C?) -- the one with a real
pretty white case, seperate monitor, fancy looking keyboard.  Maybe
begins with an F like Formula Systems or something.

2) (Multiprocessing) I think the proper model here is that one has 
several powerful (68000) processors in your personal computer, some
of which are attached to special devices (like two sharing the disk,
two sharing the screen, one doing mouse and keyboard as well as general
purpose work).

All processors would be general purpose "process pools", and programs
would be built up from processes communicating with an IPC mechanism.
In a sense, it would be simply an artifact that some of the processors
happen to be connected to the graphics display.  One would use a general-
purpose distributed processing language to organize use of the processes
available on each processor.  That is, each processor would be running
its own instantiation of the operating system, and would think of the
other processors as if they were on a tiny local net.  Everything
would communicate with message passing, although this could be optimized
with a multiple-port memory shared among processors, as opposed to a
tiny ETHERNET-style arrangement.

This would make it possible to realistically think of writing programs
which devoted themselves to a single aspect of the graphics display
which operate in parallel with other non-display programs.

I think a reasonable amount of compute power in such a configuration for
really friendly use would be a pool of 10 68000's (the updated version
of 68000, of course).

By the way, I have implemented a distributed processing language which
allows one to build programs composed of processes communicating via
message-passing.  It works with VAXes running Berkeley UNIX which are
connected via 3 MB Ethernet.

-- Lars