To: pups@minnie.cs.adfa.edu.au
Subject: Version 7 for the PERQ
Date: Tue, 24 Nov 1998 10:46:15 +0000
From: Alan F R Bain <A.F.R.Bain@dpmms.cam.ac.uk>
Message-Id: <E0ziFz6-0001EI-00@moose.dpmms.cam.ac.uk>
Sender: owner-pups@minnie.cs.adfa.oz.au
Precedence: bulk

A question which I have been meaning to ask for a while and which I 
was reminded of by the discussion of the AMD 2900 bit slice processors
was a version of AT&T V7 unix for the ICL PERQ computer.  This was so
similar to the original that the manual was an original AT&T
one with instructions on which pages to pull out and throw
away and some new ones to insert.  [basically all PDP11 hardware
specific ones go; and there are some new bits such as chatter as
simple serial comms program].  I have several binary only
distributions of this -- it was called PNX.  

What I'd be really interested to know is how it evolved from V7;
in particular the new version of `m40.s'.  In particular it seems
to run on top of a rather weird instruction set which isn't
very like that of the PDP11 (which would have seemed like an obvious
choice at first sight for a soft-microcodeable machine).  The
use of as and cc with options to write out assembler is considered
as `not a user option' in the manual; although it still works.

I have to say that in general the port seems quite bad and in
need of lots of work to make it correctly functional.  However
it's nice to have V7 readily available on a graphics workstation
with 1Mb RAM and 768x1024 display :-)

I'd be very interested to find out more of the source to PNX
or especially the microcode

Alan