Serial Number 4553
The DECSYSTEM 2020 is another processor in the pdp-10 line of 36-bit machines,
the CPU is referred to as a KS10. It was DECs last implementation of this
architecture to make it into the field (further developments were cancelled
before release). Internally, it uses a set of model-specific cpu and memory
cards, (the cpu is built out of AMD 2901 bit-slice chips) with the rest
of the system built out of conventional Unibus components - DZ11 terminal
controllers and RH11 Massbus for the disk and tape. The hardware was interesting,
but it's the magnificent operating systems - TOPS-10 and TOPS-20 - for
which these systems are best known. If you want to experience them firsthand,
visit http://www.corestore.org/tops.html - warning, it's a large download (which will take a minute or two even
on broadband) and it's a Java applet which will ask for permissions; it
IS kosher!
This particular system (which was in use until I collected it in 1993)
came from Robert Gordons University in Aberdeen, Scotland, where it ran
the main administrative system. I removed the complete system, comprising
the 2020, 2 TU45 tapes, and 5 RP06 disks. I also got many disk packs, OS
and application tapes, and complete documentation. I had to dispose of
all but one of the RP disks (to a maintenance company) due to lack of space,
and sods law means that the one I have kept has turned out to have a PSU
fault... I couldn't run it anyway as I had no 3-phase in the workshop at
the time.. I'll probably build a new system from tape onto RM03 or RM80,
but I can't get the thing to boot from tape yet.. keep getting an error
code. I must be doing something wrong, as a previous 2020 I had did exactly
the same thing... probably a flaky massbus controller on the tape or somesuch.
The current status of the system is therefore 'working' but not 'up' -
it worked ok when I switched it off at the site, and everything still checks
out on the console, but I haven't built a complete system yet. Probably
objects to a tape boot whilst no disk attached.
An archive shot of my original 2020, rescued from Chiltern Electronics
circa. 1991... this is the machine that was lost when a pipe burst in the
location where I had moved it for temporary storage. Some other bits of
early Corestore visible: to the left, a TU80. To the right, an 8/E, 11/44,
11/24, various DEC and Fujitsu disk drives, a couple of VT100s, and a VAX
11/750.
Try again... this time no floods! The last operational DEC 36-bit computer
in the UK, decomissioned 1993 - you can see one of the two TU45s to the
right.
Open up the front, it's all there. KS10 CPU & memory in the bottom,
BA11 Unibus box up top, holding the RH11 tape massbus controller, DZ11
8-port serial port cards, and printer controller.
Close-up of the CPU... Unibus interface / RH11 disk boards & console
interface in the left side of the frame, CPU in the middle, memory boards
on the right. This machine is maxed-out with 512KW of memory. That's KW
- KWords - 36-bit words remember. Think of it as a little over 2MB.
Looking in from the rear, the back of the CPU, the back of the Unibus box,
and the infamous 'Mighty-Mite' power supply which runs the KS10 part of
the box... have heard they're not the most reliable of PSUs, fortunately
I have a spare, for some reason DEC kept one on-site :-)
Close-up of the inside of the rear, above the Unibus box. Massbus connectors
for printer, disk & tape, console cable is plugged-in, KLINIK remote-diagnostic
option isn't fitted, serial lines for the DZ11s on the left side.
The QC sticker on the frame reveals this machine went through final QA
on 6/3/81
A fair number of 2020 systems have survived, they're obviously a much more
manageable beast than the KL10s!
http://starfish.rcsri.org/rcs/DECsystem/DECsystems.html the RCSRI have one.
http://www.ludd.luth.se/gallery/ludd1-mII/image03.jpg there's one in Sweden - probably several more, knowing the Swedes...
http://world.std.com/~mbg/moi_ks10.html Megan Gentry had one - she saved three!
http://www.lunar-tokyo.net/pictures/ks10/ Dan Seagraves inherited one...
http://ricm.museum.com/collections/pdp-10/2020_4224.html RICM have one, and an *excellent* webpage
http://www.panda.com/ and last but certainly not least, Mark Crispin, a hacker of high order, has three!
starfish.rcsri.org/rcs/DECsystem/FAQ/Serial_Number_Master.pdf is a list of all known systems
http://www.aracnet.com/~healyzh/pdp10emu.html quite a few people are running 10s in emulation