Obsolescence Guaranteed Newsletter

December 2024

Contents:
New projects | PiDP-10: first months | 2025: PiDP-1
PiDP-8 news | PiDP-11 news | PiDP-10 news | help with planned projects?

A lot of things have changed in the past two years. After 7 years of making these kits at the dinner table, that was no longer a realistic option. The new PiDP-10 and upcoming PiDP-1 just need too much work and space. So, we’ve set up and handed over to a dedicated little micro-factory! Giving the possibility to do many more projects. And, we can finally send out assembled & tested machines as well as kits.

At the same time, Obsolescence Guaranteed has grown to be quite a large, informal group of hobbyists. And ambitions have grown accordingly: to provide ‘fully functional computer history capsules’ (that’ll be replicas, yes) covering a wide part of computer history. Roughly, everything before the microprocessor arrived to spoil things.

New projects

The PiDP-8 came out in 2015; the 11 in 2018, with a remarkably constant stream of new builders to this day. Work on the PiDP-10 started in 2020, and after struggling with injection molding again (Development blog) it finally became a kit earlier this year. The launch party was at the MIT Museum in Boston. Very fitting, as the reconstructed ITS operating system is at the heart of MIT’s history. You'll see quite a few of the original MIT Hackers here:

mit introduction of PiDP-10, a PDP-10 replica
In early 2025, a PiDP-1 replica will appear. For 2026, a Whirlwind replica is planned. Project Whirlwind, in the 40s and 50s, resulted in the first interactive computer – to be used not with punch cards but with a display and keypad (but not an alphanumeric keyboard, interestingly). Guy Fedorkow has meticulously reconstructed Whirlwind hard- and software, and the Whirlwind 3.14 (say, Pi) will get kit builders back to the Stone Age of computing :-)

First 6 months of the PiDP-10

Already 800 PiDP-10s have been built, putting us well on track to the 1,000 needed to earn back the injection mold. We'll have our Projects Budget back for use in some new ambitious projects soonish.

The project’s entry page is here. There also are some ongoing PDP-10 projects, you’re very much invited to join, see further down.

PiDP-10 glamour shot
If you’ve only got a vague idea of what a PDP-10 was: it was DEC’s entry into mainframe-class computing. Vastly powerful, it allows dozens of terminals to be connected. But more importantly, it became a hacker playground at MIT’s AI Lab. With the meticulously reconstructed ITS operating system, there’s hundreds of games, applications, demos and languages on the supplied hard disk image. Making this the only mainframe ever that is fun to play with casually!

Coming in 2025: the PiDP-1

The PiDP-1 was done by Angelo Papenhoff (software) and Oscar Vermeulen (hardware). We really want this to be a lower-cost replica, so it uses a case made from PCB material. We hope it will attract democoders to write games and demos for the PDP-1’s interesting “Type 30” vectordot display. The dream is that this grandfather of personal computing will become something of a retro games console... why not? Vector graphics are much more fun than pixels.
With the PDP-1 team at the Computer History Museum, we plan to do an annual programming competition: the winner gets invited to run his code on the last working PDP-1 in the CHM in Mountain View. And to make sure PiDP-1 code runs exactly the same as on the real machine, the PiDP-1 uses a simulation at the circuit level, so it will be extremely exact, and still only need a low-cost Pi Zero 2. We hope it will draw a younger generation to the PDP-1 - make it live forever ;-)

PiDP-1 prototype

PiDP-8 news

New software update: The Raspberry Pi 5 is now supported thanks to Bill Cattey and Warren Young. See https://obsolescence.dev/pidp-8-quick-install.html. But the Pi 5 (even more so than the Pi 4) is so fast you better throttle the simulation speed down a notch. To make your PiDP only 2 times as fast as a PDP-8/e, add a line “set throttle 800k” to your config file (that’ll be /opt/pidp8i/share/boot/0.script for OS/8).

Steve Tockey and HB added a new cycle-exact simulator to the PiDP-8i repository. It might become the default software for the PiDP-8 soon - in the times of the Pi 2, the heavier CPU load might have been a reason not to go for Cycle Perfection, but in the days of the Pi 5 we're actually looking for things to keep the Pi's CPU a bit occupied. And a cycle-exact simulator matters for coding things like graphics demos. See posts on PiDP-8 Google Group and cycle-realistic code branch on Tangentsoft.

Update, December 9th: Work has begun on an add-on software package for the PiDP-8, to add both the GUI/terminal simulator niceties we just added to the 11, and bring the cycle-exact simulator in as a run-time option. It'll take a week or two yet.

Very useful: Ian Schofield wrote OS8Viewer, a tool that lets you view/edit/transfer files on/to the OS/8 disk image. See PDP-8: How-to-Use, in the “Mid 70s: Using floppy and hard disks in OS/8” section.

Ian also made a neat new VC8 display simulator. You can run and compile it for Windows, Linux and of course the Raspberry Pi. It will give you a display to view Spacewar on any of these. See PDP-8: How-to-Use, in the “1974: Play spacewar!” section.

PDP-8 OS-8 Viewer

PiDP-11 news

There's a major new release of the PiDP-11 software package! See the updated Building Instructions.
It includes the 'new' unix v1 (see below), can pull in the latest updates to RSX-11 and 2.11BSD (these are the oldest operating systems on the planet with active support!), and now lives on Github, so as to allow for a more active update cycle. The install process now follows that of the PiDP-10, with a pidp11control command to start/stop the simulator and a pidp11 command to grab the PDP-11 console terminal from anywhere. VT-11 graphics now work without hassle. Last but perhaps not least, Angelo Papenhoff's VT-52 terminal simulator and Rene Richarz' Tek4010 simulator are integrated into the package. The Pi's desktop ties together the eye candy (but you can still run the package on a headless Pi as well):

PiDP-11 desktop

1st Edition UNIX has been reconstructed! You can read about it here. Install on your PiDP-11 through the above update, or do a manual install:
  cd /opt/pidp11/systems:
  sudo wget obsolescence.dev/pidp11-sw/unix1.tar.gz
  tar -xvf unix1.tar.gz
  sudo nano selections
    Then, in the nano editor, insert a line “0101unix1” to make the new boot option known to the PiDP-11.
  Reboot with the front panel switches set to 101.

In further news...

  • Steven Stallion made a hardware I2C I/O board that can be driven from PDP-11 code. It’s the first bigger project that uses the I2C interface inside the P(i)DP-11. You can get the board here on Tindie.
  • The famous PDP-11 pocket reference card is very useful – and this one was made so it prints out just right on a regular printer with double-sided capability.
  • Chase Covello keeps 2.11BSD updated with the latest patches; his drop-in disk image also contains the minimal web server and weather station demo. Vlait did some BSD kernal hacking so in2.11BSD, you can access the front panel lights/switches directly.
  • 2.11BSD proves its social skills by adding this IRC client in just 300 lines of C (bring it in through ftp, or small enough to just paste into the vi editor), it compiles out of the box with cc.
  • Johnny Bilquist regularly releases radically new RSX-11 M Plus versions, here's the entry page for the PiDP-11. There’s been some notable progress, you just need to save his new disk images in /opt/pidp11/systems/rsx11mplus. Or update your PiDP-11 software package. Amazing networking things can be done with RSX-11M+!

  • Last but not least, Lars Brinkhoff has written an extremely exact simulator for the DEC VT-100. It works beautifully with the PDP-11 and the authentic display brings the proper atmosphere to any laptop. Joe Pasqua and Michael Gardi let your 3D printer make a proper physical replica if you want to go further.

    vt-100-emulator

    PiDP-10 news

    Unusual for mainstream media, the Guardian in the UK ran an article on the PiDP-10, getting the word out on preserving computer history. And CuriousMarc made a very nice Youtube movie of the PiDP-10, ITS, and their history. See it here.

    Shrdlu was the first-ever AI that inhabits a 3D world full with objects it is aware of and can manipulate, in response to an English language conversation. The demo has been around for a long time – but that, in fact, was just a screen recording (which the ITS operating system easily lets you do).
    In September, Eric Swenson (@eswenson) edited the original old LISP and reconstructed the MicroPlanner code from the early 1970s it relied on, to run on the latest version of Maclisp. This reconstruction was thought impossible for decades, it’s a major achievement. Getting this alive on a PiDP-10 is a nice and manageable way to learn about Lisp on ITS too. It's also a major step in preserving the earliest history of AI.

    shrdlu on the PiDP-10

    Muddle and MIGS (Muddle Interactive Graphics System) have been resurrected. It is worth exploring, just for seeing what was possible in 1973 already:

    muddle-MIGS on the PiDP-10

    Terry Kennedy went all-out with a PDP-10 cluster; mounting additional Pis on the back of the PiDP-10 so he's got TOPS-10 and TOPS-20 to connect to from the PiDP-10 itself. Which gets you thinking - you'd could do an ARPANET replication mounted to the back panel. All with a single USB-C cable (delivering power and networking either through Wifi or through USB Gadget mode); the Pi 5 power supply can handle all of this through the Pi 5's USB ports :-)

    muddle-MIGS on the PiDP-10

    Join to help with these planned projects!

    Maybe these are potential topics to kick off on the PiDP-10 and PiDP-11 Google groups:

    Porting to microcontrollers and FPGAs:

    Although I think most of us rather like the Raspberry Pi, there's no reason you could not drive a PiDP-8 front panel through a humble $3 microcontroller. In fact, an esp32 is all you need for a perfectly networked 2.11BSD PDP-11 - with the work of spritetm. I tried, and this $11 LILYGO TTGO T8 V1.8 ESP32-Wrover board plus a $2 SD card is literally all you need for a full PDP-11! It lives happily networked to my PiDP-11. if you don't like telnetting in to it, add 4 wires and a tiny TFT display, plus Bluetooth keyboard.
    muddle-MIGS on the PiDP-10
    Because Spritetm ported simh to the esp32, it would not even be too hard to let it drive the PiDP-11 front panel directly. Project.


    PDP-10 networking setup and reconstructing the ARPANET:


    IMP processor
    The PiDP-10 is very much capable of not just networking (see supdup, the local Pi 5 is already networked into the PDP-10 by default over Chaosnet) but going out on the internet, and HECnet is another logical thing to connect to.
    But with the IMP simulation functional, a project to recreate the 1973-or-so ARPANET would be very worthwhile. The hardware can live on the back of a PiDP-10 in the shape of a bunch of Pi Zeros. Weaving in some P(i)DP-11s would get to recreate a good part of ARPANET. Would be a nice project indeed! But a first step is to see if anyone has the source code for the IMP from that time. It appears to be lost, so far.

    Update, December 9th:
    Lars Brinkhoff has an exciting correction to the above. The 1973 IMP code has been recovered, and brought up in simh. As he writes:

    "That is not so! David Walden et al did a good job recovering it and Bob Armstrong (Spare Time Gizmos) made it run on SIMH. There is a sample configuration to set up a network of 3-4 IMPs. Charles Anthony (Multics emulator etc) made an IMP-host interface; I debugged it and integrated it with SIMH. I made some tests running a network with a Linux NCP standing in as hosts; interfacing with ITS, WAITS, etc can also be done.[..]"
    https://walden-family.com/impcode/
    https://github.com/open-simh/simh/tree/master/H316/tests/
    https://github.com/larsbrinkhoff/linux-ncp/

    Update 2:
    I think this is interesting to many. Because it looks like resurrecting a working version of Arpanet is feasible.
    From a conversation with Lars:
    " >> So - enough of the IMP is reconstructed to have a couple of IMPs connected via virtual serial connections? It's just that the ITS code to work with the IMP has been lost, is that it?

    No, the ITS code has not been lost. First, the MIT archive has ITS sources from 1971 to 1990, so it's possible to get things from there. [..] Here's an overview of the situation for all operating systems: https://gunkies.org/wiki/Network_Control_Program_(ARPANET)

    >> So an ARPANET with PDP-10s and PDP-11s is close...

    The major obstacle is that the 1974 IMP code that provably works, only allows the "short leaders" used in the original IMP-host interface. Some time mid-70s a newer "long leaders" protocol was introduced. The ITS version we use now have "long leaders" and is hence incompatible (why of course) with the IMP code. So I think the solution would be to convert back and forth where necessary to have the IMPs believe it's still 1974. I started on some Arpanet testing and automated scripting for bringing up a network. [..] https://github.com/larsbrinkhoff/linux-ncp

    I intend to put a converter in between to translate between the two formats. The "long" version is mostly a superset, and for a while both ran in parallel. So as long as we restrict the network to 63 IMPs and 4 hosts on each, it should work. The converter will be transparent, so neither the IMP nor the ITS code would have to change. "
    Obsolescence PiDP Newsletter Dec24

    To be continued...


    Tweets

    Photos


    Web-Stat traffic analysis