Subject: Re: VAX now runs multicpu!
To: Carlini, Antonio <Antonio.Carlini@riverstonenet.com>
From: Thor Lancelot Simon <tls@rek.tjls.com>
List: port-vax
Date: 05/30/2001 19:29:49
On Wed, May 30, 2001 at 03:40:01PM -0700, Carlini, Antonio wrote:
> 
> > Thor Lancelot Simon wrote:
> > 
> 	>I think you have the model numbers mixed up.  A 785 is a 780 with
> an
> 	>upgraded processor board.  A 782 is two 780s connected by a
> shared-memory
> 
> 	This is all correct.
> 
> 	>box.  Though I don't think they were ever officially cataloged, the
> 	>additional ports on the shared-memory box could be used to build
> "783"
> 	>and "784" configurations and reportedly at least some of these were
> in
> 	>fact shipped from the factory as that configuration and had actual
> DEC
> 	>VAX 11/784" badges on them (there was one such at Ontario Hydro,
> IIRC).
> 
> 	I've never seen anything that suggested that there was
> 	ever a machine called the VAX-11/784 (or 783 or 787 etc.)
> 	There are plenty of web pages out there that claim that Mach 
> 	was developed on a 784 (amongst other things). But most of
> 	these pages are obviously copies of the "master" page and I
> 	don't see anything at all authoritative on those pages.

I'm pretty sure the 784 existed, though not for long.  I recall finding
the part number for the "11/784" badge in one of the big thick white
books when I worked at DEC many years ago.  I've also spoken to people
who said that they'd run into 3- and 4- processor systems in the field
that were running VMS in asymmetric multiprocessing mode using the MA780,
and one person in particular I met at a long-ago conference who proudly
said that *his* site had one of the few machines with an actual DEC 11/784
badge.  As I said, I think that was Ontario Hydro.  Also, I think the
multiprocessor PDP-11 they had was an 11/70M (a Massbus PDP-11); the 11/74
is just a Qbus J11 machine with a Unibus behind a bridge; it was general
wisdom in the late -11 era that swapping a 70 for an 84 (the 74 is just a
slower 84), as DEC liked to suggest, was often not a good idea because of
the lost I/O bandwidth.  Or are you talking about a multiprocessor J11
machine?  That'd be interesting.

I note that http://telnet.hu/hamster/vax/e_1984.html *mentions* the
11/784 though there's definitely other bogus info on that page.  I'll try
to dig up a better reference.

The DEC office I worked at supported a MA780-and-dual-785 machine at one
of our more important customer sites.  The 785s had been upgraded from
780s but still had their "11/782" badges on them; the customer kept
joking about how they wanted the correct badge for their 11/787 but I
don't know if that was an official product designation or not, nor for
that matter whether any 11/785+MA780+11/785 configurations were sold
as new units, rather than as upgrades-in-place. :-)

> 	There is a paper floating around that describes the
> 	steps some group went to to build their own
> 	multiprocessor VAX by disassembling a few 780s.
> 	I forget if they went the whole hog and put more than
> 	two cpus together.

This is the paper linked to from George Goble's home page that describes
a machine that replaces one of the other system bus boards in a 780 with
a second CPU, isn't it?

-- 
Thor Lancelot Simon	                                      tls@rek.tjls.com
    And now he couldn't remember when this passion had flown, leaving him so
  foolish and bewildered and astray: can any man?
						   William Styron