Foonly

~ the obituary ~

At the start of the 1970s the Stanford Artificial Intelligence Laboratory wanted to replace its DEC PDP-10 KA10 with something far faster, and got DARPA money to study building it. The plan was ambitious in a specific way: rather than adopt TENEX, which was becoming the ARPANET standard, SAIL intended to leapfrog straight from the timesharing system it was running to a new generation of both machine and operating system. ARPA cut the funding for both halves in 1974. Most of the design team went to DEC instead, where they contributed heavily to the PDP-10 KL10.

Dave Poole was one of the principal designers and he did not let it go. He incorporated Foonly Inc. in Mountain View, California, on 7 June 1976, and built the machine: the F-1, using emitter-coupled logic gates, faster and hungrier than the parts the architecture usually sat on. It was the fastest PDP-10 ever built. Exactly one was made.

Building it emptied the company. Foonly spent the early 1980s selling smaller and much cheaper PDP-10 compatibles instead, the F2 and F4 and F4B and F5, and Tymshare tried marketing the line to its own customers as the Tymshare XX Series. None of that recovered what the F-1 had cost.

~ the case file ~

~ the name ~

The name is an error message. A PDP-10 assembler at SAIL printed FOO NLI when handed something it could not use as a symbol, meaning that FOO is not a legal identifier, and the laboratory named its hypothetical successor machine the Super Foonly out of that. Poole then took the joke onto a California business license.

~ one machine ~

The F-1 ended up at Information International Inc. in Culver City, better known as Triple-I, where it became the computational engine behind some of the computer-generated imagery in Tron in 1982 and in Looker the year before. It carried around two megabytes of core memory on an AMPEX unit and talked to its peripherals over F-bus, a Foonly design, because DEC's own Massbus would have tied the machine to DEC's own disks.

Omnibus Computer Graphics of Toronto bought it from Triple-I in 1984 with plans to be the first large digital animation house in the film business, and ran it on television network logos for the CBC and CTV and on Flight of the Navigator. Dave Sieg, who looked after it there, described it as a fascinating research project that had never been designed for the spares and expertise a production shop needs, built on five machine-wrapped wire-wrap pages. The only part of this story that ever reached me was the movie, and it reached me as a film about computers rather than as a film made on one.

~ foonex ~

The cheap machines had a software problem that no price could fix. They ran Foonex, a TENEX variant, instead of the TOPS-20 that PDP-10 customers actually wanted, which narrowed the market to people willing to port their lives. What shipped were wire-wrapped engineering prototypes, so every site needed unusually competent staff, and reliability suffered accordingly.

The Jargon File, which is where most of what is written down about this company lives, is blunt about the rest: Poole's temper was legendary and his tolerance for fools was not. It also records the parrot that sat on his shoulder and went everywhere with him, which is the detail everyone who met him seems to have kept.

~ the mars ~

DEC canceled its Jupiter project in 1983, leaving the PDP-10 world without a next machine, and Foonly proposed building another F-1 to fill the hole. Systems Concepts won that argument with its Mars, and Foonly never quite recovered. The California record closes the company on 19 April 1989, by which point the architecture it had spent thirteen years serving was finished anyway.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »