I thought I’d dig a bit further back into my personal history and take a look at the Intelligent Filestore Project.
This project started at Manchester University and was taken to The University of Essex by it’s head, Prof. Simon Lavington. It’s aim was to make prolog-database-style knowledge management efficient by building a highly parallel hardware search engine.
The project won a couple of awards…a Silver medal from the BCS (before my time, when the project was still at Manchester Uni.) and a Bronze award from the Alvey Directorate (the people paying the bills).
The kit had a hardware “Lexical Token Converter” (a hashtable for symbols), a main controller CPU and 128 custom designed and built search processors, each with 128Mb of RAM (this was the mid-80’s, remember).
You can see this hardware:
It isn’t me in the picture… Actually, I had more hair in those days.
This picture shows two of the systems: the dev/test rack and a finished half-height cabinet (at the right). This unit was eventually shipped off to the University of Edinburgh’s Artificial Intelligence Applications Institute.
My job involved adapting a bog-standard Sun 3/75 VME-bus based workstation to act as the controller for the specialised hardware. It’s in the picture, on the half-height rack. This work was interesting…we even got the schematics for the box from Sun (under an NDA) and eventually found a bug in their hardware. Of course, they found lots of bugs in ours…
I also found myself doing a lot of general Sysadmin-y stuff on a variety of unix-based kit (including SunOS boxes, Microvaxen, High level Haradware ‘Orion’ , GEC-63 stuff). This was also a lot of fun. I cut and sharpened my unix/sysadmin teeth here, and *I* think they remain sharp to this day!
I also played a lot of sport: badminton, squash, tennis. This accounts for the wreckage that is my shoulder these days.
I learned a few useful things on this project:
- All hardware must include 1+ 7-segment LEDs to that you know what the hell is going on.
- It is possible to write one academic paper per decade. All one then needs to do is reshuffle the paragraphs and resubmit…it’ll get published somewhere…
- Any sufficiently advanced technology is indistinguishable from magic (Clarke’s 3rd law); “sufficiently advanced” is not the same as “truly advanced” and is very much according to the eye of the beholder.
- “Thou shalt not work with team ‘mates’ that are continually getting up and coming down.” Nefarious substances can do strange things to people and it is really hard to get a team to ship product when the team doesn’t know what day it is or what state they may have left their wife in as a result of last night’s binge. The threat of physical violence in the workplace is not conducive to one’s best work…
- If a system needs a reboot, people will complain irrespective of how regularly the system is rebooted or how much advance notice was given. Ignore the complainers and remind them that the rest of the world generally wakes somewhat before noon.
- “I just found out…” is not the same as “I just performed world-class research.” Unless you are applying for a research grant, when it is.
- Just because it has always been done “that way” doesn’t mean that it can only be done “that way.”
As an example of the latter bit of wisdom: the Manchester Uni. team had always performed their IPLs (Initial Program Loads) by shipping S-record encoded binaries via a serial port interface on the main board; a slow and error prone process. This is what they had always done. It was what everybody did. Not knowing that, I came along and wrote a 10-line ‘ C’ loader program that ran on a daughter board and simply mapped in the appropriate memory and pushed the requisite bits into the mapped-in segment. Immeasurably faster and easier…
Here’s a little secret: the project was demonstrating to “The Minister” for Science and the hardware just wasn’t working! The software-only simulator was, so we just demoed that, all the while running a diagnostic tool to ensure that the actual hardware’s status lights were flashing happily. The Minister asked a question: “Is it fault-tolerant, as well?” One of my colleagues said “Of Course!” and reached down into the rack and pulled out a known-dud board. The simulator (of course) didn’t miss a beat and the status lights on all the remaining boards kept flashing happily. “Amazing!” said The Minister and walked away, very impressed.
Happy days!
Now an exercise for you, gentle reader: go back and brush up on Clarke’s 3rd law, it you still haven’t understood it…
For all it’s ups and downs, this was a good job. I made friends that I still keep in touch with, I learned a LOT, had a good time and at the end, it was a stepping stone for me to get a lecturer’s post in the University of East Asia, Macau.
One can’t really ask for more from a job, can one!
