[ On Sunday, December 26, 1999 at 08:37:23 (-0600), Jay Maynard wrote: ]
> Subject: Re: NCR Tower ?
>
> The Tower 32 is a 68020-based machine from about 1987 that normally runs
> System V release 3, pretty much straight as it came from AT&T.

If you're lucky -- many ran a very early port with lots of old r2-isms
kept in user-land because the NCR folks were overly concerned about
backwards compatability.

> It's at the
> slow end of things by today's standard. IIRC, it used the stock 68020 VM
> hardware. It's probably got no more than 16 meg of RAM and a QIC-150 tape
> drive on a dedicated controller; the disks are probably MFM or RLL, maybe
> first-generation SCSI.

sounds about right....

I ran one that supported about 20 or so people using it as a DBMS server
(custom app written in FoxBase and/or some BASIC implementation) as well
as word processing (with WordPerfect).  We had a lot more memory in it,
and we had SCSI-II disks too.  (System disks were ST-506, with only
support for two such drives)

It eventually got replaced by a much bigger and faster 3B2 (/500, IIRC),
with the DBMS application re-implemented from scratch.  They may have
since migrated to PCs for all I know.

> Hardware docs are likely to be nonexistent, since the
> machine was built before NCR abandoned the 68000 line for Intel, which in
> turn was before AT&T bought them out and then spun them off again.

The add-on cards were straight Multibus-I (IEEE-796), and reasonably
stock components as well (though likely designed by NCR).  There was
special firmware for the SCSI controller though, IIRC, and it had to
boot from one of the main ST-506 drives (unless I'm thinking of a
different Motorola box).

> I strongly doubt any NetBSD will run on it without a porting effort that's
> likely to be quite difficult; OTOH, you should have the tools you need to
> make such an effort right on that system. In its time, compiling stuff off
> the net for the Tower series was pretty straightforward, since there were no
> significant weirdnesses to the Tower's Unix.

Indeed.

I did have tons of freeware running on the machine I managed.

I happen to have kept a "Tower-32 Hardware Service" manual (32-Bit
version, October 1985, D1-0518-A), but it's not too interesting from an
O/S implementor's point of view.  It gives the wiring diagrams for the
power, internal cabling, etc., as well as pinouts, jumper settings,
etc. for most everything (though little detail on the CPU, memory
expansion boards, and mass storage controller).  It does document the
*usage* of the "start-up subsystem" (boot-ROM) with some very minimal
details of the memory map and how the "hot/warm/cold-start" stuff works.

I can answer any questions this manual might provide me the details for,
or I could offer to copy of any of the most interesting pages if
necessary.  I suppose I could even lend the entire manual out, but I
don't really think it's worth it unless you're needing to service the
machine to get it actually running first.

I can probably even obtain one or more machines myself, though I've no
idea how I'd find the time or incentive to help with a port though!  :-)

> I'd recommend leaving it as a stock SysVR3, myself, unless you somehow
> manage to lay hands on hardware docs for it.

It would be very cool to have Multibus support in NetBSD!  :-)

There were zillions of these machines made and many are likely still
running POS systems right up until the then of this year....

Who knows -- maybe it even needs a new TOD clock chip too...

(i.e. it will probably require y2k patches...   which reminds me -- I
should pull my 3b2 out of the garage next year and see if it still runs
too!  :-)

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>