stacken.txt

                             Stacken Computer Club

                            by Jan Michael Rynning

Back in 1978 Nils Soderman invited a group of micro computer enthusiasts to a
meeting at his home. He wanted to set up a computer club, and it was agreed
that the Royal Institute of Technology was the best place to recruit members to
the club.

One day, on my way home from the institute, one of my course-mates told me he
had seen posters announcing that a computer club would hold its first meeting
that evening. He knew I was interested in computers. I turned back and just got
there in time for the meeting. As usual I couldn't keep my big mouth shut, and
was elected vice-chairman of the club, which was named the Stacken
Microcomputer Club. ``Stacken'' is the Swedish for ``the stack''.

During the first year the major activity of the club was to give lectures on
how a microcomputer worked, and how to build one. A microcomputer kit was
constructed. A number of members got together, and bought components for the
kit and UCSD PASCAL systems at a club discount. Discussions arose on whether it
was really neccessary to buy 64 Kbyte of memory and two floppy drives. The UCSD
system would run in 48 Kwords, with only one floppy, though that was not
recommended. Most settled for 64 K and two drives.

In 1972 QZ, the Stockholm University Computer Center, had purchased its first
DECsystem-10, with a KI10 CPU, and 96 Kwords of core memory. Originally it
was only intended to be used for research, but in 1977 the Royal Institute
of Technology started using off-hour time for education. Two years later,
in 1979, the Department of Numerical Analysis and Computing Science bought
a DEC-2020, running TOPS-10, to be used for research and education, instead
of renting time from the computer center. Members of the computer club
persuaded head of department Yngve Sundblad to adopt the policy of letting
the students use up all unused computer time. The club was given a PPN, which
was used by all its members.

With the arrival of practically unlimited access to a large computer, the club
attracted many new members, whose main interest was programming, not building
their own hardware. The main direction of the club thus turned from hardware to
software, and the ``Micro'' was dropped from the name, to reflect that the club
was no longer confined to small machines.

A lot of software projects were carried out throughout the following years, the
most notable being the AMIS project, supervised by later chairman Lars-Henrik
Eriksson. AMIS, which doesn't stand for anything and has a dozen
interpratations, is a screen editor modeled after Richard M. Stallman's EMACS.
It's written in PASCAL, except for the low level I/O routines, which are
written in MACRO-10. Byte shuffling and the low level search routine are also
written in assembler, for efficiency. As soon as the first version was up and
running on TOPS-10, club members managing the PDP-11/45 educational computer,
running RSTS/E, at the School of Electrical Engeneering asked for a copy, to
port it. The interface to the system dependent routines proved to be well
chosen, but the PASCAL code required a number of changes, due to compiler
differences and PDP-11 integers being only 16 bits. After that we set up a
standard for how the PASCAL code should be written to be portable. The text
buffer handling had to be rewritten from scratch. On the PDP-10 we had used the
large address space to house the entire file being edited, but on the PDP-11
there was no room for that. In fact, AMIS had to be heavily overlaid to fit
into the small address space of the PDP-11, using a specially written overlay
handler, and a LISP program to optimize the overlay structure. AMIS was later
ported to VAX/VMS, PRIME and Norsk Data machines.

In 1982 DEC salesman Hans Petzelius asked Stacken if we wanted an old
DECsystem-10, which was going to be scrapped, unless we accepted it. Of course
we wanted a mainframe of our own! It was a KA10 based system, immediately
dubbed KATIA, ``TIA'' being the Swedish for ``TEN'' (and KATIA being a girl's
name). The institute promised us a computer room, but the decision of which
room to give us got stuck in the bureaucracy, and the machine was parked in a
garage.

It wasn't until the end of 1983, when Peter Lothberg came to the club, that
things really started to happen. Peter is the most persistent person I've ever
met, and by the end of 1984 the bureaucrats finally signed the paper giving us
a 900 sq. ft. room in a basement at the far end of the institute.

In the meantime Peter had managed to lay hands upon QZ's old KI10 (by that time
QZ had replaced it with a dual-processor KL10 system), rented a computer room,
and installed it, assisted by other members of the club. That machine was named
KICKI (also being a girl's name, though it should have been spelled KIKI,
reflecting the ambition to upgrade it to a dual-processor system). After
negotiations with QZ Peter was allowed to connect it to the ANF-10 network, so
it could be reached from the terminals at the institute.

The machine was dismounted and moved to Peter's computer room on Thursday and
Friday. Monday morning a DEC serviceman said to another ``He'll never get that
machine up and running'' (when QZ soldit to the Swedish Pulp and Paper Research
Institute in 1977, it took DEC over a month to get it running steadily) and got
the answer ``I logged in on it an hour ago''. After that noone questioned
Peter's hardware competence.

KATIA and its periherals were taken to Peter's computer room, to use KICKI for
debugging it. The KA10 CPU cables were found to be missing.  Fortunately, the
Medical Computer Center at the University of Gothenburg had had a KA10 (it now
has two KI10's), which had been scrapped some years earlier, and for some
reason they had kept the cables. After about a week's work we had both the KA10
and 1 1/2 MF10's in a working condition, and started running the CPU tests. As
soon as the machine entered user mode, it got a illegal memory reference. The
relocation and protection power supply circuit boards were also missing. The
fast registers and the relocation and protection are the only parts of the KA10
CPU built out of TTL, and the ground level of their power supplies is adjusted
to convert the 0 and +5 V TTL lebels to the ordinary -3 and 0 V, used by the
rest of the CPU logic. I suggested that we use the fast register power supply
as a replacement, as the machine can use core instead of fast registers, but
Peter promptly installed a laboratory power supply, and adjusted it to the
right levels. After that it ran. Not using fast registers would have degraded
performance by some 30-40%. Later, the Medical Computer Center dug out the
missing circuit boards, and admitted they probably had a complete KA10 save
cabinets stowed away.

Running TOPS-10 6.03A on the KA10 wasn't much of an attraction, as we already
had a KI10 running 7.02. We knew of ITS, the Incompatible Timesharing System
developed at the MIT AI Lab, and of their home built paging boxes that extended
the limited KA10 physical address space of 256 Kwords, so we wrote them a
letter asking them if we could have ITS and a copy of the prints for the paging
box. In return, we received a box of 7 channel magnetic tapes. We only had 9
channel drives, and noone we inquired knew of anyone who had a 7 channel drive,
except for the MIT AI Lab. Again, the Medical Computer Center came to our
assistance. When the Max Plack Institute in Berlin discontinued its three KI10
systems, the Medical Computer Center drove down with two trucks, and picked up
as much as they could. Among that was a 7 channel drive, which they took by
mistake. As they had no use for it, we could have it.

A few months later, Peter and another member of the club drove down to
Gothenburg to pick up equipment for which they had no room. There were a KI10,
to make KICKI a dual-processor system, 22-bit channels (we only had 18-bit
channels), two TU40's, more memory, and some other goodies. On the way back the
other guy fell asleep at the steering-wheel, the truck went off the road, and
landed on its side on a field. The cargo was spread out all over the place. The
driver of the wrecking truck that came to their rescue wanted to use his crane
to remove the cargo to lift the truck, but Peter finally convinced him the
cargo was more valuable than the truck. Give-away KI10's don't grow on trees.
As time went by most things were put back into a working condition. The CPU bay
1 and bay 2 cabinets were askew, so two straight cabinets from MF10's with
damaged backplanes were sacrificed, and the CPU back planes, cabling, and power
suplies were moved to those cabinets.

In the fall of 1985, one year after we got KICKI up and running, at last we got
the electricity installed in the computer room at the institute, six months
after schedule, and moved KATIA there. It took us a few days after we moved it
to get it back running. The problem that was the hardest to find was a cold
soldered short circuit that had been there all the time, but never before
caused any trouble.

KATIA was then inagurated by Gunnar Brodin, then president of the institute,
now university chancellor, who was a great support the the club in its struggle
with the bureaucracy. As DEC couldn't give the machine away, I, then president
of the club, presented one Swedish crown (ca. 15 cents), the agreed upon price
of the machine.

Right after that Peter set off for the University of California Irvine, to pick
up a KI10 system, which Mike Iglesias had saved from being sold to a junk
dealer. Mike couldn't accept that the machine he had cared for for a decade
were to be sold to a junk man, so he sent out a message on the ARPAnet, asking
if someone wanted to buy it. We didn't want to see it scrapped, and we had use
for it, so we offered to pay the scrap value.

The people who helped us and all the bits and pieces of hardware and software
we received, built and wrote over the next years are too numerous to mention.
DEC has been very helpful with otherwise-to-be-scrapped hardware and software.
We also owe many thanks to the Royal Institute of Technology, especially to the
Department of Numerical Analysis and Computing Science.

As of this writing, KATIA has a BBN paging box and runs TENEX 1.35. KICKI is a
tri-SMP KI10 running TOPS-10 7.02 with TCP/IP and DECnet phase IV over DDP
devices, and IBM disks and tapes on a System Concepts SA10. The D/A converter
is used for playing music and for the acoustic part of the cuckoo-clock built
from a discarded disk drive. The Cuckoo-clock real-time program is started
automatically once an hour by the BIGBEN routine. The PDP-8 and PDP-15 were put
in moth-balls to leave room for a PDP-11/70 running RSTS/E. Another PDP-11/70
running RSX is used as a DECnet junction for the other machines. We also have
two KS10's, alternately running ITS, TOPS-20, and test versions of TOPS-10.
We've been promised a 2000 sq. ft. house on the campus, for our KL10's, in
addition to the old computer room and our club room.