The Pegasus 1 was first delivered in 1956 and the Pegasus 2 was delivered in 1959. Ferranti sold twenty-six copies of the Pegasus 1 and twelve copies of the Pegasus 2, making it Ferranti's most popular valve (vacuum tube) computer.
At least two Pegasus machines survive, one in The Science Museum, London and one in The Manchester Museum of Science and Industry. The Pegasus in The Science Museum ran its first program in December 1959 and is regularly demonstrated. It was the oldest working digital electronic computer in the world until 2012, when the restoration of the Harwell computer was completed.
- The necessity for optimum programming (favoured by Alan Turing) was to be minimised, "because it tended to become a time-wasting intellectual hobby of the programmers";
- The needs of the programmer were to be a governing factor in selecting the instruction set; and
- It was to be cheap and reliable.
The first objective was only partially met: because both program and the data on which it was to operate had to be in the 56 words of primary storage, it was often necessary to resort to tricks in order to reduce the number of transfers between that store and the drum memory. To what extent the third objective was reached, depends on how one views a price of £50,000 for Pegasus 1 without tape drives, line printer or punched card input and output, which required an hour or more of preventative maintenance by a resident engineer every morning, before a programmer or operator was allowed near it.
Pegasus had eight accumulators, seven of which could also be used as index registers. (It was the first computer to allow this dual use.) Accumulators 6 and 7 were known as p and q and were involved in multiply and divide and some double length shift instructions. It had 56 words of fast memory stored in nickel delay lines, which was supplemented by a magnetic drum holding 5120 words. A word was 40 bits, of which one bit was for parity checking. Two 19-bit instructions were packed into one word and the extra bit (not counting the parity bit) could be used to indicate a breakpoint (optional stop), to assist in debugging. It had a relatively generous instruction set for a computer of its time, but there was no explicit hardware provision for handling either characters or floating point numbers.
The speed of arithmetic operations was about the same as the Elliott 402 computer, which could add in 204 microseconds and multiply in 3366 microseconds. The Pegasus basic instruction cycle time for add/subtract/move and logical instructions was 128 microseconds. Multiply, divide, justify and shift instructions took a variable time to complete. Transfers to and from magnetic drum were synchronous and had to be optimised where possible. The layout of blocks on the magnetic drum was interleaved to allow some processing between transfers to/from consecutive blocks.
In 1956 the first Pegasus was used to calculate the stresses and strains in the tail plane of the Saunders-Roe SR.53; the results were used to check the manufacturers figures; the programmer was Anne Robson. Because of the importance of a computer it was housed in the drawing room, complete with an Adam's ceiling, of Ferranti's London office in Portland Place.
In 1957, a Pegasus computer was used to calculate 7480 digits of pi, a record at the time.
- Early British Computers, by Simon Lavington, Digital Press (US) and Manchester University Press (UK), 1980, ISBN 0-932376-08-8.
- The Pegasus Story: A history of a vintage British computer, by Simon Lavington, London, Science Museum, 2000. ISBN 1-900747-40-5.
|Wikimedia Commons has media related to Ferranti Pegasus.|