LEO was advanced for its time, as it was one of the first computers which could be programmed to handle a series of different operations and tasks without the need for rewiring the computer itself. Rather, as mentioned it ran using different programs, created specially by the likes of David Caminer. Obviously, LEO had another advantage commercially as it was the only computer of its time created solely for business purposes.
Rather than simply list the achievements of LEO and the different computers of their time and compare them in that way, this section looks at the actual specifications of each computer where possible, comparing them between each other to show which was the most technologically advanced.
Firstly, we should look at the computers from which Lyons took their inspiration for LEO: ENIAC, EDVAC, and EDSAC.
Starting at the beginning of the digital computing age as we now know it, we have ENIAC. This computer was developed by the United States Military in July 1943, and was to become the first of a number of computing ventures undertaken as a direct result of World War II. ENIAC was composed of vacuum tubes, crystal diodes, and relays. It was used specifically for mathematical calculation, such as projecting the trajectory of artillery to be used by the army. In terms of operations, ENIAC was capable of approximately 385 multiplication operations, or 40 division operations per second. Timings on the amount of equations the ENIAC could perform per second depended on the size of the operation it was being requested to calculate. Smaller operations would run faster, whilst larger sums would take fractionally longer to compute, however, as this timing was no longer than a second or two, this was not considered to be something of a problem.
ENIAC was a modular computer, with each module or panel designed to perform a different function. There were a number of these panels, with many of them serving the computer for arithmetic uses, whilst the others were used for powering and shutting down the ENIAC (Initiating unit), Synchronising the other units (Cycling unit), printing the results from each operation using an IBM card punch (Printer unit), reading the information from punched cards (Reader unit), a Constant Transmitter, three Function Tables, and of course a unit which was the Master Programmer. This final panel was the one which controlled operations from the other panels during operation. One of the problems with ENIAC was that for a different program to operate on the machine, you couldn’t simply insert a new program. Instead, for each operation of ENIAC which differed from what it was currently running, the machine would have to be rewired. This was a time consuming affair which would then delay the new operations.
Unfortunately for ENIAC, its Achilles heel came in the form of the vacuum tubes it was made from. During the start up and shut down process for ENIAC, the heat generated by the computer put these valves under a great deal of stress. Many of the valves within ENIAC would break, causing the machine to break down. This problem initially led to ENIAC being offline around half of the time it was expected to be operational. Eventually, a decision was made that ENIAC should not be turned off when not in use, as this would reduce the thermal pressures that were breaking the vacuum tubes. This decision led to faults on ENIAC being reduced to no more than a handful of vacuum tubes breaking every few days. An interview conducted with "Pres" Eckert and presented by Computerworld in 2006 revealed that “a tube failed about every two days and we could locate the problem within 15 minutes. We invented a scheme to build the computer on removable chassis -- plug-in components -- so when tubes failed we could swap them out in seconds.”
Operations on ENIAC eventually ceased on the 2nd October 1955, when ENIAC was switched off for the final time.
The successor to ENIAC is often considered to be the EDVAC computer. This was designed by the creators of ENIAC John Mauchly and J. Presper Eckert even before the ENIAC computer was fully operational. Once again, the computer was, like ENIAC before it constructed at the University of Pennsylvania's Moore School of Electrical Engineering for the U.S. Army's Ballistics Research Laboratory and was conceived to calculate munitions range tables.
EDVAC (Electronic Discrete Variable Automatic Computer) was designed to be an improvement in almost every way to ENIAC, resulting in faster, more efficient operations. Many of the improvements made to the system came as a direct result of problems encountered during the creation of the earlier ENIAC machine, or through moments of inspiration which would have improved the earlier design.
The final design differs greatly from ENIAC, as EDVAC consisted of a magnetic tape reader-recorder, a control unit with an oscilloscope, a dispatcher unit used to receive instructions from the control and memory and direct them to other units, a computational unit to perform arithmetic operations on a pair of numbers at a time and send the result to memory after checking on a duplicate unit, a timer, a dual memory unit consisting of two sets of 64 mercury acoustic delay lines, and three temporary tanks.
The new architecture of the EDVAC was based on what became known as the Von Neumann Architecture. Although design began on the EDVAC in 1944, the machine was not delivered to the Ballistics Research Laboratory until 1949. Even then, the machine still had a number of obstacles to overcome before it begun operations in 1951. Even then, due to a legal dispute, EDVAC was only running operations on a limited basis. Over the years of its life, EDVAC received a number of upgrades and improvements as technology allowed, some of these included including a new punch-card input/output device, increased memory storage in the form of magnetic drums, and a floating point arithmetic unit. Unlike ENIAC, EDVAC ran operations for a longer period of time, with much fewer faults.
By the time EDVAC ceased operations in 1961, the number of computers built by universities in America and around the world had started to increase dramatically. The Ballistics Research Laboratory had themselves ordered another computer, the ORDVAC which ran along side the EDVAC performing similar operations throughout the 1950’s before both machines were retired and succeeded by BRLESC.
The main problem which beset EDVAC was not one relating to the computer itself, but rather the design of the machine. The University of Pennsylvania and the EDVAC designers Eckert and Mauchly had a dispute over the patent regarding the computer due to its use of Von Neumann Architecture. The argument was that the publication of the first draft report on the EDVAC publication prevented the patent of the machine. This dispute eventually led Von Neumann to be credited with the creation of EDVAC rather than Eckert and Mauchly, who claimed that much of the information in the report predated the arrival of Von Neumann at the university. Subsequently, the two men left the University of Pennsylvania as a result, to form their own company.
Unfortunately for all parties involved, it was this dispute which led to another computer being named as the World’s first electronic stored program computer. In England, the Manchester Mark I ran its first successful operations in April 1949.
This computer, the first to be built in England, took the credit by virtue of running throughout the time in which the heated arguments caused by the Von Neumann draft. The Mark 1 was similar in architecture to the computers built across the Atlantic; however, reaction to the computer in England was rather more positive and widespread than that of its American cousins. Newspapers such as The Times wrote articles about the fascinating new ‘Electronic Brain’ which had been constructed in Manchester. Running only between 1949 and 1950, it ran a number of mathematical operations, however, rather than have these used for tasks such as calculating munitions tables, the Mark 1’s operations were purely for the purpose of the Manchester University. These included a search for Mersenne primes, as well as investigatory work into the Riemann hypothesis.
The Manchester Mark 1 caused a great deal of fascination in the academic realms of England, and soon afterwards, another computer, this time built in Cambridge University using the Von Neumann report as a basis for its construction. EDSAC, or Electronic Delay Storage Automatic Calculator was built by Maurice Wilkes and consisted of mercury delay lines for and vacuum tubes, much like other computers of the time. Input for EDSAC punched tape and output was via a teleprinter. Like other such computer built by Universities, both in England and in America, EDSAC was used for mathematical purposes relating to research undertaken by the institutions. However, the EDSAC itself was important not for the work it conducted for Cambridge University, but for the role the computer and its developers had in the construction of another computer soon after EDSAC’s creation – the LEO I.
Back in America, things had improved for Eckert and Mauchly. Their company, Electronic Control Company grew rapidly, with staff who had worked in the EDVAC project with the men joining the company having left the university. Small scale contracts had been won to build some small calculating devices, but Pres Eckert and John Mauchly had much bigger plans, and announced the construction of the World’s first commercial computer. Due to the fact that this new machine was intended for universal use across any number of companies who would by them, it was fittingly called UNIVAC.
This computer would not be required to calculate scientific calculations, instead merely operating simpler arithmetic functions and data transport. UNIVAC was designed to use a new type of input, rather than the traditionally used punched cards. This was to be an advantage to UNIVAC, and a hindrance. Using this magnetic tape allowed faster operating times, as different punched cards would not have to be fed into the computer, however when the computer was designed, the technology did not exist. Rather Eckert and Mauchly had to develop their own magnetic tape system from scratch, which increased design and production time, as well as increasing the cost of the build exponentially. Later, once the computer was operational, the use of this tape held up some initial sales of the UNIVAC computers, due to the fact that many of the intended target customers for the computer had a great deal of information held on punched cards. From their point of view, if there was no way to transfer the information to the new computer, then they were effectively useless to their business. This problem was rectified by the introduction of two new products from the company – the UNIVAC card to tape converter which would translate the card information using a unityper, and the UNIVAC tape to card converters which would perform the reverse operation for the output.
The downfall of UNIVAC, and the Electronic Control Company came from the fact that most of the money which had been used to research and build the computer had come from loans. In the mid to late 1950’s, this money had to be paid back to the creditors. As this amount was unavailable to return, the company was sold to Remington Rand. Mismanagement meant that whilst 41 of the UNIVAC computers were sold to various companies and organisations, sales of UNIVAC never reached their full potential.
Each of these computers with the exception of the then top secret Colossus, were key in one way or another to the birth of computing, and played one of the biggest roles in the decision to create a computer to deal with operations such as Payroll for J. Lyons Co.
In every term, LEO was an improvement of colossal scale compared to machines such as ENIAC and EDSAC. The creation of the Manchester Mark I and the work at Cambridge University were of course obvious influences on the decision to create LEO, and whilst technically a descendant of EDSAC through following the same Von Neumann architecture, it was innovations by the Lyons team in particular which set LEO ahead of everything which had come before. It is interesting to note that LEO had a counterpart in America at the same during the same time which actually had an advantage over the Lyons machine. The UNIVAC team had successfully managed to create a working version of the tape drive which Caminer had originally envisioned for LEO. Apart from this, the abilities of the two computers were very similar. The key difference being that whilst the UNIVAC team had access to funds from a number of sources, some stemming from the BINAC project, the LEO team had worked on what has been said by David Caminer himself to have been a shoe string, making whatever they were occupied with at the time work.
It is safe to assume that had they survived long enough to compete, the UNIVAC team and Electronic Control Company itself would have stood little chance in competing with IBM when they finally arrived in the computing business. Instead, it is likely that the manufacturing and marketing problems which beset LEO Computers would have similarly affected Electronic Control Company.