Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!bbn!rochester!ritcv!tropix!mjl
From: mjl@tropix.UUCP (Mike Lutz)
Newsgroups: comp.arch
Subject: Re: QM-1 or like machines???
Message-ID: <288@tropix.UUCP>
Date: Sat, 5-Sep-87 12:10:02 EDT
Article-I.D.: tropix.288
Posted: Sat Sep  5 12:10:02 1987
Date-Received: Sun, 6-Sep-87 06:13:31 EDT
References: <63900006@convex> <43700020@uicsrd>
Reply-To: mjl@tropix.UUCP (Mike Lutz)
Organization: GCA/Tropel Div. Rochester, NY 14450
Lines: 73

In article <43700020@uicsrd> lawrie@uicsrd.csrd.uiuc.edu writes:
>
>>>>/* Written  3:35 pm  Aug 13, 1987 by graham@convex.UUCP
>>>>There was once a company named Nanodata which built a machine called the
>>>>QM-1.  The QM-1 had two levels of "micro-code" (one called nano-code and the
>>>>other call micro-code).
>
>Burroughs, AKA Unisys, designed a machine called the "D" machine,
>around the early to mid ''70s.  We had a prototype here at Illinois.
>It had a nanomemory. ...  I'm not sure if it ever made
>it to the marketplace--in my memory I associate it with the 1800;
>that was either it's final name or the machine that won out over it.

Ok, the straight poop (from an old poop) about the QM-1, the D-machine, and
the Burroughs 1700/1800.

1. The QM-1 grew out of a research project at the University of Buffalo
   in the late 60's, early 70's -- Project Mu, headed by Bob Rosin.
   I was a lowly, junior, graduate student assigned to the project.
   We were investigating the uses of user-level microprogramming:
   experimental computer architectures, high level language support, etc.
   After looking at the D-machine, a Microdata processor, and a couple
   others, we decided none were suitable, so some folks from
   UB's engineering lab formed Nanodata and proposed what became the
   QM-1.  The main advance of the QM-1 over the D-machine was that
   vertical microinstructions were interpreted by a *sequence* of
   horizontal nanoinstructions: the opcode of the microinstruction
   gave the address of the first nanoinstruction, and other fields
   in the microinstruction were accessible by the nanocode to select
   registers and ALU operations.  On the D-machine, the
   microinstruction selected a single nano-instruction, and no
   parameterization was provided.

2. For technical and political reasons, Project Mu never took delivery
   of a QM-1 (though Nanodata thoughtfully provided summer employment
   for many of us students :-).  Instead, we purchased a B1700 in
   1972; the B1700 was a technical marvel for the time:
	a) multiprogrammed emulation, with different microcode for
	   each language in use (FORTRAN, COBOL, PASCAL, etc.),
	b) a multitasking, virtual memory operating system that fit
	   in 32Kbytes (you could run applications with as little as
	   48K of main memory; we went wild and purchased 64K),
	c) bit addressable main memory, making it a snap to emulate
	   existing computers, or to try some real exotic architectures,
	f) a variable precision ALU (1-24 bits).
	e) and, as they say in marketing, "a whole lot more".

   The B1800 was a successor this machine; as I understand it, the
   low-end B700/B800 lines were based on repackaged D-machines.
   From personal experience, I will state that the B1700 was a
   pleasure to work with at the microcode level (and anyone who has
   done serious microprogramming knows what an amazing statement
   that is!)  While not a "RISC" machine, the B1700 was optimized for
   emulation, and the pieces just fit together well.  It was difficult
   to write bad code, and it always seemed that information ended up in
   the right place for further processing.

Mike Lutz
GCA/Tropel
tropix!mjl

P.S.  At the time of Project Mu, dynamically modifying the underlying microcode
   was viewed as a "neat idea": one could optimize performance of particular
   applications on the fly, etc.  The idea has pretty much been abandoned,
   as the potential for mass confusion overwhelmed the actual benefits.
   Oh, well.

P.P.S.  Anyone wanting more information on the state of microprogramming in
   the early to mid 70's should take a peek at "Fundamentals of
   Microprogramming" by Agrawala and Rauscher (Academic Press, I believe).
   They have a good section on the QM-1.  For the B1700/B1800, look
   for a book by Elliott Organick and Jim Hinds, titled something like
   "The B1800 Architecture", also by Academic Press, I believe.