World’s oldest original working digital computer gets a reboot

Ron Czapala

Computing has come a long way. Take a look at this clip of the Harwell Dekatron, otherwise known as the Wolverhampton Instrument for Teaching Computation (WITCH). The supercomputer from 1951 was restored over a period of three years by experts at England's National Museum of Computing in Bletchley Park. This week, they rebooted the beast in front of reporters.

And what a beast it is. The world's oldest original working digital computer is the size of a garage door and resembles something from a campy science-fiction flick about Martian invaders. True its functionality is rather modest by today's standards (it can multiply two numbers in less than 10 seconds, for example), but darned if it isn't impressive to watch. Lights blink, tubes whir and switches flicker.

More info: http://en.wikipedia.org/wiki/Harwell_computer

localroger

WITCH wasn't a supercomputer in its day; as the name suggests it was a teaching computer, meant to give students experience but not necessarily practical levels of performance. It's entirely electromechanical and therefore very slow even for an era when practical electronic computers were using mercury delay lines and vacuum tubes. To practical computers of its era WITCH was more like a programmable calculator or microprocessor trainer. But this is also probably why it is the oldest functioning computer; the basis for its design is much more straightforward than the sometimes obscure electronic hacks that were used in those early days, when it was common for real computers to work well in the morning after fresh maintenance but become unreliable in the afternoon.

Mike Green

Some of these early computers / calculating machines had unique idiosyncracies. A mentor of mine in college had gone to college at Princeton where he got a part-time job sitting in front of the IAS machine watching the Williams tube memories for a sudden bloom of light due to a programming error. Apparently, the incrementer for the the program counter was triggered by a bit in the instruction word. If it was miscoded, the machine would repeatedly fetch the same instruction word every instruction cycle. Since the Williams tubes used a destructive read mechanism and would have to rewrite the instruction, tubes with a 1-bit would have the same location rewritten repeatedly. A few seconds of that and the phosphor coating on the inside of the tube would vaporize very nicely leaving that bit permanently zeroed. His job was to hit the emergency stop button when there was any sign of this. There was a chart on the wall showing what bits of what words were already damaged in this way so they could be avoided in writing programs or storing data.

kwinn

Hardy pioneers indeed.

LoopyByteloose

You know, all those tubes emit X-rays off the top ends. I actually built an X-ray machine once with a design out of The Amateur Scientist in Scientific American.

I wonder how much of a radiation burn you could get from that device.

lanternfish

localroger wrote: »

... when it was common for real computers to work well in the morning after fresh maintenance but become unreliable in the afternoon.

I know people like that!

Heater.

Of course prior to these tube, or even relay gadgets, "real computers" were actually humans calculating huge tables of functions. That was what their job was called "computer". As we are talking about Britain here they probably had to stop for tea at 3..00pm:)

Heater.

Loopy,

You know, all those tubes emit X-rays off the top ends

I don't think they do. In order get any kind of meaningfull X-rays out of a tube you are going to need to put 10 or 20KV across it. Here is a nice account of doing just that http://www.belljar.net/xray.htm

That particular computer is using dektron tubes that light up between 400 and 600 volts. My bet is that they are X-ray free, no more dangerous than a neon lamp. Normal computer triodes probably run at about 200v so no worries there I think. Besides dekatrons don't realy have a "top end" like a thermionic valve.

I might be more worried about sitting next to a neon display sign in a bar which do seem to have power supplies up in the 10 and 20Kv range.