Sir Maurice Wilkes centenary - 'Flash-Gordon' tech

History of computing at Cambridge

As well as Wilkes 100th anniversary, this year is also the 75th anniversary of the Cambridge Computer Lab that he set up ( and for which EDSAC was built). I came across an interesting history of the the Cambridge Lab published this year. It's a relatively non-technical history of the early days as well as the rapid developments from the 1980s onwards.

"Cambridge Computing' by Prof Haroon Ahmed

http://shop.tmiltd.com/shop/home/pId/176

Valves are mostly used today (outside of audio) where you need high power AND very high frequencies (radar, satcoms, particle accelerators), or switching insane amounts of power quickly (e.g. the EEV CX2025X thyratron can switch 100kV at 15kA peak (1.5GW) in tens of nanoseconds).

Most people now have a valve in their home, the magnetron in a microwave oven (also originally a WW2 radar technology).

This all very well but..

I thought that the Manchester 'baby' was the first electronic stored-program computer in June 1948 and led to the Mk1 and the Ferranti Mk1 ( first commercially available computer). There's a plaque on the wall recording the event near Manchester Dental Hospital

My physics teacher had worked on the Manchester computers and had a photo of himself, stripped to the waist, working in a sweltering nightmare of racks holding chassis. The 'baby' was apparently 17 feet (5.2 m) in length, 7 feet 4 inches (2.24 m) tall, and weighed almost 1 long ton (1.0 t).

As a by-note I started in electronics at ~13 years old with valves, first a regenerative radio and then a record player with a valve amplifier

Staggering Diversity

The speed of change in the early 1960s was staggering.

The English Electric (Computer Division) regularly changed its name with each new merger/acquisition. Having extended to "English Electric Leo Marconi" by 1966 - the next addition of parts of Elliot Automation was resolved by "English Electric Computers". In 1968 it became ICL when merged by Government dictate with the equally acquisitive ICT.

In 1967 - before that final merger - English Electric Computers had the following systems on its Kidsgrove site:

A DEUCE was used for the company payroll. That was valves and mercury delay lines. Subject to problems when the overnight cleaners unplugged the mercury delay line heaters in order to use a vacuum cleaner.

The Bureau service was processing Share Registration and Min Ag Fish punched card data on a KDF8 (RCA 501). Discrete transistor logic - with a Friden Flexowriter for console output. Operator decisions were input on a large panel of illuminated coloured buttons. That met visitors' expectations of what a computer should look like - especially when the room lights were dimmed in the evening. You could tell an operator by the callouses on their thumbs from rapid button settings. The magnetic data on the tapes could be read by eye with a little device effectively containing iron filings.

The Bureau service also had an English Electric "home-made" KDF9 doing scientific work like Rolls Royce turbine calculations. Still using transistors - but the operator could now input commands on the Friden Flexowriter. There was even a hard disk and the Egdon system could timeshare several programs. The papertape reader was so fast that static electricity could build up in an unearthed basket - and a spark would reset the cpu.

In the next Bureau bay was a state-of-the-art imported RCA Spectra 70/45 - an IBM 360 compatible machine using integrated circuits. The replaceable disks were 8MB. One of its main workloads was to develop the operating systems for the System 4 range of computers. The proposed range of machines was 4/10, 4/20, 4/30, 4/40, 4,50, 4/60, 4/70, 4/90 - each from a different hardware team from several of the merged companies. Similarly there were teams working on separate operating systems based on papertape/cards, magnetic tapes, or disk systems eg 1D, 3D, 3J, 5E, 5J, 7J.

In the development area was the prototype 4/70 with the massive memory of a total of 1MB in several six feet high cabinets. The CDC fixed disk unit weighed one and a half tons - and had water cooled bearings. The platters were at least two feet in diameter. There were two vertical stacks of platters - with the head armatures linked so that they counter-balanced the inertia forces as they moved in and out. The total capacity was a mammoth 600MB.

There was also a KDF7 for industrial control. One stood outside rusting in the rain after a fire on a customer site. When powered up its first output was the time and date of the failure. It had a large Ni-Cd

battery - with large nuts on the terminals. These were close enough to give rise to the apocryphal story of a engineer's spanner getting welded between them.

Analogue computing was done by a KDN2 (?).

The LEO 3 production line was elsewhere.

...and that is only a sample of the concurrent diversity that was haemorrhaging money in the British computer industry.

I thought the pun was unnecessary "Flowers, who overcame resistance from colleagues,", but an interesting article. On the subject of tubes, they disappeared much more recently than the article would indicate - CRT's are, of course, valves :), and there are probably still some of those in use in some places...

EDSAC is also 100 this year

If you are the sort of person who realizes that thumbs are not fingers.

In a more serious vein, Manchester Baby was more a "proof of concept", while EDSAC was intended from the get-go as a serious tool. Not to take away from either, but get 10 history buffs in one room and you'll probably get 11 opinions on "First computer", depending on the qualifiers.

And... "UNIVAC" used tubes/valves. The (later) "Univac Solid State" computer, a.k.a SS80 and SS90 (as used by Don Knuth and Grace Hopper, among others) was the one with the diode-core logic (and, yes, 6 power tetrodes on the "clock", and over 50 Thyratrons in the printer). It was also the first computer I modded, although not the first I ever programmed.