EMAS on 2966 (was Re: Xenix on the Acorn 32016)
gtoal at gtoal.com
Sat Oct 22 22:18:35 CDT 2005
Jules Richardson <julesrichardsonuk at yahoo.co.uk> wrote:
> Graham Toal wrote:
> Now that's interesting. Having seen some of Acorn's internal source and
> the way they often worked, it seemed there was a lot of "just go ahead
> and do it" philosophy going on. Maybe someone mentioned 32000 Xenix in
> passing and then some other programmer took it upon themselves to add
> support in to the linker :)
The only programmer was Mark Taunton (also Ex Edinburgh, hence why the linker
was written in Imp, and also why the link format had a lot in common with
that of EMAS!). I think he may still be working for ARM somewhere. I bet
he'd remember. I had a listing of it because I was the only other Imp
programmer there and I was helping find a storage bug that was causing
link errors for some program of mine. (The listing has a few of my
> Again, I'll check that I don't have it on disk when I get home in a month...
It was written in Imp. I'm sure you'd remember if you'd seen it!
> > The Edinburgh collection is getting quite impressive. We also have
> > recovered locally written operating systems for the PDP9 and PDP15,
> > two O/Ses for PDP11s, Perkin Elmer 32bit, 68000s and of course the
> > big one, EMAS for ICL29XX
> That last one caught my eye, given that we have an ICL 2966 which we'd
> like to (try to) get operational one day (no mean feat, given how
> dismantled, rusty, and spread all to hell and back across lots of
> storerooms it is at the moment!).
> Seems like there are "lots" (as in more than one :) of people/places who
> know about the 1900 line, but the 29xx's seem to be largely forgotten.
> Of course a complete 1900 would be nice, but I doubt there are any left
Who is 'we' in this context? I can't imagine anyone just happening to
have an ICL mainframe lying around, never mind the power to run it or
the manpower to put it all back together again!
(Hold on a sec... quick visit to Google... aha! so you're the
curator at Bletchley Park! That explains a lot :-) Hadn't seen
your Retrobeep site before! If I were still in the UK, I'd be happy
to drop by with a shitload of Acorn hardware I grabbed before it headed
for the dumpster)
So yes, it would be fantastic to get EMAS on line again, and if your
hardware works, it's definitely doable because the 2900 EMAS is the
only one we actually have full binaries for! And the University
has pretty complete archives on a CD jukebox where we could go fetch
any binaries that we're missing.
What's especially lucky is that we have a boot tape, which you're
welcome to download and see how far it gets you:
We have several people on the history project who might be
interested in a project like this; If you think you might be serious,
I would encourage you to join our mailing list and tell folks what
you're thinking of and ask for volunters. Some are retired; two
unfortunately work in the US. I'll put together a list of the
people who would be relevant to this and we can contact them
individually if you decide to start up a project. I would lay odds
that if you could get funding and it became obvious that it was a
serious project, we could get help from Edinburgh University (EUCS
which used to be called ERCC) and pull in a lot more folks than just
the ones already in the History Group.
We don't have binaries for either the ICL4/75 version or any of the
later IBM mainframe (or NEC/Fujistsu/Hitachi clone) versions, so although
emulators exist for the IBM (Hercules), there would have been a major
bootstrapping problem in cross-compiling the sources to IBM binaries.
And although we had ICL2900 binaries, we have no ICL2900 emulator.
It never occurred to me that we might find 2900 hardware somewhere. This
could be a major break for our project if you decide to run with it
and the hardware can be made serviceable. Also a working system would
be the perfect place to start from to write a 2900 emulator.
The biggest obstacle that I can think of would be building a disk
image with all the relevant files on it. We have them as individual
files, not as a disk image. But we have *major* amounts of source
code and even an Imp compiler, so we ought to be able to hack the
disk drivers around and create a user-level program to build a disk
Incidentally, I discovered an ICL engineer on the net today, who worked
on the EMAS hardware:
- I'll be dropping him an email this evening to tell him about the
Edinburgh project. If you start up a 2900 project, he sounds like
someone well worth speaking to.
Let's talk more about this, either here or offline, or preferably
on the Edinburgh group at
PS Does Bletchley have any systems that can read DECtapes? We still
have a few tapes in our project from the PDP9 and others which we've
never read back in. Including one for an operating system called
"DECsys" which I believe some of the guys on this list are desperate
More information about the cctalk