Yes. The 8 & 9 machines were ECL, the rest were TTL. IIRC, those were the
32/87, 9780, PN9600.
David mentioned disks on the PN (Unix) series. Those were formatted with
multiple of 512 byte sectors. The RTM/MPX machines used 768 byte sectors,
which was super optimal for the disks they happened to ship with their
earliest machines, but then a right PITA for everyone who used the machines
for decades beyond that. It was not just the strange size, but I think the
minimal disk allocation unit was something like 16K, and you only got 8 or
16 chances to add new segments to that. You better know how big your file
would grow before you started writing.
It's all slowly leaking back into my brain. The early machines, were number
32/xx and ran RTM, their Real Time Monitor. The xx was, IIRC, 27, 75, 77,
87. Very much process control oriented. A terminal *could* be hooked up to
an editor task that could edit code and submit jobs, but then it could not
detach and let you interact with another program. The company I worked for
hacked up a task swapping capability on top of it so we could actually get
work done. That was 1976 or so.
Around 1982?, they added an MMU and introduced MPX, the Mapped Programming
Executive. That was much more usable, but still with the problematic disk
layout. I think the numbering changed then to x7nn, where X was the overall
technology and nn was a size within that. I know there was an 8750, 8780,
9780, 6780, 2750, and 7750. Those were the MPX machines. The UTX (Unix)
machines replaced the 7 with a 0, giving David his PN9080 and PN6050.
For unrelated reasons, I have to clean my basement today. Who knows what I
will dig up.
On Sat, Oct 15, 2016 at 1:07 PM, ANDY HOLT <andy.h...@tesco.net> wrote:
> ----- Original Message -----
> > From: "David Brownlee" <a...@absd.org>
> > We had a PN9080 and PN6040 at City University as the main systems in the
> > late 90's
> Ah, yes, "The Magic Roundabout" - was three 6040s and one 9080. I still
> possess the Gould nameplate from the 9000.
> They were the last machines we had that we though of as mainframes (even
> if many would call them minis - but I think the racks were wider than 19"
> so they clearly weren't minis!)
> There's some interest in the story of how this system came together:
> we did have a Honeywell dual 66/60 which was supplemented by the 9000 as
> a time-sharing system when we had had it 5 years.
> After two more years we calculated that we could buy and maintain the
> trio of 6000s for less than the maintenance cost for the three years
> that the Honeywell was due to remain and gain a noiceable increase in
> computing power (and a noticeable decrease in power consumption) by
> doing so … and actually managed to convince the bean counters of this.
> > (accessed via the usual mix of ADM3As, ADM5s, some Sun3s and a
> > whole bunch of Whitechapel MG-1s, ans some colour terminals of which I
> > cannot recall the name, but I remember them having a setting where they
> > would auto colour characters based on their clas - alpha one colour,
> > numbers another, and two or three other colours for the rest of ASCII)
> I also forget what those colour terminals were. The first Sun came along
> when it turned out that it was cheaper to buy it and an Ada* compiler than
> the Ada for the Honeywell.
> * Computer Science /insisted/ they needed an Ada compiler.
> They never used it - but the Sun was useful :-)
> > When the CS department finally moved away from their own 6040 it was left
> > forgotten in a room over the summer - in the autumn the aircon was found
> > have failed, overflowed and the machine was sitting there with water all
> > over the floor and in a steam bath. Still running fine.
> Don't remember that - but certainly believable
> > Quite robust that ECL :)
> Um, I don't think the 6000s were ECL - think they were a reimplementation
> the 9000 using cheaper technology - probably whatever was the current
> "state of the art" TTL (don't think CMOS had taken-over for speed yet)
> Thanks, Abs, for reminding me of those times.