[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: A working PDP-11 (almost ... )

>Yes, it's a J-11 CPU. Turns out our ATEX system uses dozens of them in
>parallel. Looks more like a VAX than a PDP-11, but boy does it smell the

COOL!!!  After C.mmp and Cm* at CMU and Andromeda, it is great to know that
multiprocessor PDP-11 systems were actually put into real use!

The semiconductor group once wanted to build a 1GIP hypercube system out of
1024 J-11s; Bill Strecker's group wanted to build a shared memory system
with 64 J-11s and a (snoopy?) cache; and people who worked on the first VAX
workstation (2 tubes with M68000 display processors and fiber optic lines
to an 11/780 -- it wasn't a major success except I think that X-windows got
started there) wanted to build some other shared memory system with
M68000s.  Gordon Bell held a "cycles for the masses" woods meeting.  He had
a fit over the several different hardware proposals, complaining about all
of the researches in switches they funded at the NSF, and rightfully giving
me grief for not having enough evidence that the PDP-11 address space for
each node would be enough for doing logic simulation, and decided to go
with just the Strecker's group's system but make the CPUs be MicroVAX chips
instead of J-11s -- Andromeda -- and to just go ahead and make it a product
instead of a limited-run experimental system.  But he left the company and
the poor machine had to wait for MicroVAX chips to happen and got delayed
and bounced around several times and I think they only built one.  (They
did build at least one.  There was as a monitoring board for every CPU
board, and an Ethernet local inside the machine to connect the monitoring
boards.  I wonder if they built any more for the government?)  I did some
work on a logic simulator for it but without the hardware being there yet,
I went back to working on the behavior modeling language compiler.  I don't
recall if it was before or after this that the group got a Zycad system --
special-purpose hardware for logic simulation.  I didn't work on hooking
DECSIM up to the Zycad system, but some people did an amazing job of
pulling it off.

- Aron