FR80 Technical Papers

John Rushby

Atlas Computer Laboratory, Chilton, UK, 1974-1975

Recollections

I worked at the UK Atlas Computer Laboratory from 1974-75, mostly on systems programming for a III (Triple-I) FR80 microfilm recorder. This was a then-state-of-the-art vector graphics device with a 16k x 16k raster. Various cameras, including a 35mm movie camera, could be mounted above the tube (changing them required a fork-lift). Here's a picture of me (backup, and another, in color, but full of dust specs) at the operators/programmers teletype; the camera is in the open bay to my left and the tape drives are to my right: that's not an ash tray to my right, it's a spool of half-inch mag tape. The small screen just to my left gives a preview of what is written to the main tube and the lights and switches below are for reading and modifying the contents of registers. This page shows the camera bay in the top picture and the fork lift in the second. A $100 ink jet printer today would do about as well as this £200,000 machine (about a million pounds in today's money) did then. Incidentally, III (Information International Inc.) was an interesting company (see this Wikipedia entry). They owned the one-and-only SuperFoonly F1, which was used to create a number of pioneering computer animations (e.g., in the movie "TRON"). During my first few years at SRI, our mainframe computers were a pair of Foonly F4s. (There are some recollections of Foonly and its founder, David Poole, here; I remember the parrot.)

Here is another picture of me ( backup) looking on admiringly while my boss, Bob Hopgood, signs to officially accept the FR80. Bob has created a great web site that collects a lot of material from the Atlas Laboratory. Among the material is a set of technical papers on the software we developed for the FR80; a good many of them were written or co-authored by me (e.g., numbers 2, 4, 8, 9, 11). You can even find our quarterly progress reports online (I was involved from Q3 1974 to Q3 1975; I'm identified by my initials JMR). It all provides a rather interesting window back into the world of programming in the 1970s. The FR80 was unusually primitive, even by the standards of the day (it was, after all, just a printer): its III-15 processor was a copy of the PDP15 (an 18-bit architecture) with just 4,096 words of memory, and assembler was the only means of programming it.

One progress report (Q1 1975) describes dismantling a Tektronix storage tube to find the source of catastrophic interrupts at the PDP15 (we had a genuine PDP15 with a VT04 graphics display and spark pen tablet, and a massive 1 Mbyte hard disk). Such hardware tinkering was quite common in those days. I recall another incident where tapes written with a certain software system on one computer just could not be read on the FR80 despite the hardware and software supposedly being compatible in every way. I was introduced to "MagSee" which was a bottle containing fine metal in a liquid suspension: you dip the tape in the MagSee and then, under a lens, you can see the actual bits on the tape. That revealed the problem: the filemarks had wrong parity!

On another occasion, we were debugging a program on the FR80. It was common to single-step the program at the console and inspect register contents on the lights (that's what all those switches and bulbs visible in the pictures are for). At some point, we concluded that the only explanation for some of the behavior we were seeing was that one of the console light bulbs must have blown. The FR80 had a bulb test switch, and that indicated all the bulbs were working. Many frustrating hours later, we diagnosed that that was all it tested: the bulbs were working fine, but the circuit that drove one of them from its bit in the register had failed!

Much the most interesting application that made use of the FR80 was the Antics computer animation system developed by Alan Kitching. This was a Fortran program that simulated the kind of 2-D (cel) animation used in studios like Disney. It had a rather sophisticated kind of linear in-betweening (now called morphing) and was used to generate a number of interesting movie segments. The test strip we used in debugging (featuring a DNA sequence) now runs in a loop at the Science Museum in South Kensington, London; it gave me quite a shock when I walked into the museum a few years ago and saw the familiar images. The original Antics was a distant precursor to modern computer animation systems; it has evolved greatly over the years and continues to be developed by Antics Workshop.

A few years before all this (that would have been in 1969 or 1970), I had a summer job at the Atlas Lab at a time when the Atlas computer was still there (by 1974 it had been replaced by an ICL 1906A). Atlas was the most powerful computer of its day, and physically it may have been the largest machine ever built (it occupied two floors of a large building). But the Palm PDA in my pocket today probably has more power (and certainly more memory) than the massive Atlas. The Atlas was developed at Manchester University, where I subsequently worked from 1975 to 1979.


Having trouble reading our papers?
Return to John Rushby's bibliography page
Return to the Formal Methods Program home page
Return to the Computer Science Laboratory home page