EDSAC I, 1947 P.J.Farmer R.Piggott M.V.Wilkes W.Renwick S.A.Barton G.J.Stevens J.M.Bennett (posed photograph) Copyright (c) University of Cambridge Computer Laboratory, 1999EDSAC 99 - Fiftieth anniversary of the EDSAC Computer

John M. Bennett

INTRODUCTION

On Thursday 15 and Friday 16 April 1999, the University of Cambridge Computer Laboratory celebrated the fiftieth anniversary of the launching of EDSAC 1.  (The first logbook entry was dated May 6 1949, although EDSAC had carried out useful calculations for a few days before that date.)

With over 400 attendees, the event took place at the Laboratory.  The highlights as listed in a commemorative booklet available from the Laboratory, were the following.

On Thursday evening a reception was held in the University Combination Room. This was followed by a dinner in St John's College, at which the speaker was Professor Sir Peter Swinnerton-Dyer.  Also about 90 reminiscences from Laboratory members past and present were collected.

The brochure and a two-tape videoset which includes the material summarised above is available from the Laboratory.

SUMMARY OF TALKS

The introductory talk on the first day was given by Professor Robin Milner, Head of the Laboratory since 1996.  He outlined the University Computing Service, computer science research and interactive teaching activities.

Millner was followed by Professor Maurice Wilkes, who pioneered it all.  After early experimental research on the propagation of waves in the ionosphere, in 1945 he became Head of the Mathematical Laboratory, the Computer Laboratory's predecessor.  In 1946, after attending lectures at the Moore School, University of Pennsylvania, he initiated the building of EDSAC using ultrasonic mercury delay line memories.  His pioneering work on programming with Wheeler and Gill led to the first book on the subject and was followed by his work on microprogramming, cache stores, time sharing and in 1975 (with Needham) multiple access facilities and networks - leading to the Cambridge Ring.  After retirement in 1980, he joined DEC, and is now a staff consultant with AT&T in Cambridge.  His talk described how he planned EDSAC 1 on his return from the Moore School.

Professor David Wheeler followed Wilkes.  David led the EDSAC 1 programming effort, and used it to compute the wiring schedule for EDSAC 2, which he planned after a period with the ILLIAC group at the University of Illinois..  EDSAC 2 was a microprogrammed machine using a bit slice design logic with ferrite cores, magnetic tapes, and - with an enlarged store - an Autocode.  Two Nobel prizes were awarded for research carried out with the help of EDSAC 2.

Supporting work done on machine architecture was a group of about a dozen engineers, most of whom had previous industrial and wartime (e.g., radar) experience with electronics.  Without their teamwork - illustrated by
anecdotes - the Laboratory's accomplishments would not have been possible.

The next computer project (in the sixties) was Titan, headed by Professor Roger Needham.  From 1980 to 1995, he was Head of Department, and now is Managing Director of Microsoft in Cambridge.  Titan was the last machine built in Cambridge to provide a computing service., and embodied a number of innovative ideas in time sharing and multiple access.  Begun on a shoestring financially, with the help of a substantial 1965 Science Research Council grant it began working in 1967.  Among developments which came from it were mechanical CAD developments made possible by the Laboratory's operating software.

The next talk, by Dr David Hartley  (now Executive Director of the Cambridge Crystallographic Centre) concerned providing a computing service for the University.  Providing a computing service and teaching computer science as an academic subject were two hand-in-glove activities until about 1970, when the two were separated and the service side was transformed by the availability of one or more computers to every academic.

The first talk on the second day was by Professor Andy Hopper of the
University Engineering Department, and was concerned with the design of high speed communication rings.  The first Cambridge Ring was begun in 1974 and provided 10 Mbps communication between 100 devices.  Faster rings (100 Mbps and, in the Cambridge Backbone Ring, 1 Gbps) followed.  These embodied VLSI design and
used formal design verification techniques.  The large available bandwidth provided stimulus for such applications as wirewrap machines, laser printers and various video applications.

The next talk was by the Laboratory's Professor Michael Gordon on formal (mathematical) verification of computer design.  Now, in the late nineties, this early work seems very simple.  Current emphasis is on fragments of computers. However there are signs that the verification of whole processors will again become the order of the day.

The changing role of teaching was next discussed by Dr Frank King, a
lecturer in the Laboratory.  Until the late 1960s, extradepartmental teaching consisted of having students attend selected Diploma lectures.  Television was introduced in 1968 and the trend now is away from programming to application packages.

ASD System's Dr Stewart Lang's talk was concerned with input and output through interactive computer displays.  He traced the changes from hard copy to soft copy over the last 50 years.

The use of random eye iris patterns for recognising individuals was next described by the Laboratory's Dr John Daugman.  There is little doubt that this technique is the identification technique of the future.

Ongoing research projects in the Laboratory were then summarised by Professor Ian Leslie.  He described how profound changes now taking place in man-machine interaction techniques emphasise the convergence of communication and computation.  And the concluding talk by Dr Karen Sparck Jones, a Reader in
the Laboratory, emphasised the rich variety of the Laboratory's research.

EXHIBITION

The exhibits ranged from a Millionaire calculating machine through EDSAC 1 items (mercury delay lines, etc.), EDSAC 2 and Titan units and Cambridge Ring items.  It included three EDSAC 1 simulators,  current activities of 15 research groups and details of the computing service and its networking infrastucture.

AUTHOR'S NOTE

The author's connection with EDSAC 1 may be of interest.  He joined the Laboratory in September 1947 as  Maurice Wilkes's first research student.  He designed and built EDSAC 1's main control unit and its initial orders input (bootstrap) facility.  In addition to using the machine for various technical calculations, in July 1949 he devised what was probably the first interpretive programming scheme - primarily to save storage space!

15 October 1999

See Also:

Copyright © John M. Bennett 1999
Draft V0.1 19 October 1999 - Web Markup by Tom Worthington Comments to: webmaster@tomw.net.au