Mechanical Data Processing and Calculating

First Generation Electronic Computers (1943-1955)

The development of fully electronic computers to replace mechanical calculators began during World War II 77 with development at the University of Pennsylvania of the ENIAC processor, followed later by the EDVAC. Work on the project began in 1943 and went into service in 1946 to build an electronic computer able to do ballistics and atomic energy calculations. By 1948 ENIAC had been modified run from stored programs (Weik, 1961). EDVAC was designed from the beginning to use von Neumann's architecture78. Capabilities to process data into information then began to grow by orders of magnitude each decade. Although the first experimental electronic computers were used for military and scientific calculations79, the technology soon began to be applied to accounting and management requirements of large organizations.

Remington Rand first began developing commercial electromechanical computers in the US market, with the little known 409, under Gen. Leslie Groves's leadership (during WWII Groves led the Manhattan Project to develop the atom bomb - Fay 1996; Wenning 1997, ????). The first products of this line of development were delivered to their first customers in 1952.

Remington Rand's first business computer, UNIVAC I, followed on from the ENIAC and EDVAC work begun at University of Pennsylvania by J.W. Mauchly and J.P. Eckert. They both resigned from the University of Pennsylvania in 1946 as the result of a dispute over who owned patents for the technology, and went on to form formed their own company, which won a contract to build a computer for the US Census bureau. Remington Rand bought the company in 1950 as Mauchley and Eckert were running out of money, and funded completion of the development and production (Weston 1997; Gray, G. 2001). The first UNIVAC was delivered to the US Census Bureau in 1951. Major innovations in the UNIVAC were use of mercury (acoustic) delay lines to provide processor memory and metalic magnetic tapes (solid metal ribbons!) for storage. The more than fifteen UNIVAC I (Figure 8) systems implemented by 1955 were installed by defense organizations, two insurance companies, two railroads, and a university80.

As a general rule, [in the service bureau] individual research, engineering and mathematical projects have numerically exceeded straight data processing jobs while the greater overall volume of machine time is devoted to the latter. In order to keep programming costs at a minimum, extensive use is made of the Library of Univac I Routines whenever possible.... Business applications such as payroll reporting, cost account reporting, sales statistical summarizations and various statistical analyses have been done for a number of firms. Scientific applications include the engineering problem solutions from areas such as helicopter design, nuclear reactor design, bearing design, geodetic surveys and many others... [T]he two systems are operated back-to-back applied to insurance activities. [Picture and quote from Weik (1961a).]

Figure 8. UNIVAC I installation at Franklin Life Insurance Company, Springfield, Ill. Franklin operated a second UNIVAC I as a service bureau. Staff for the two systems included 3 supervisors, 32 technical staff (analyst, programmers, operators and service technicians) and 50 clerks (presumably key-punch operators)! The price for a basic UNIVAC I system was $950,000, including the central computer with power supply, supervisory control desk, and 10 Uniservos tape drives - where a 1,500 foot magnetic tape could store up to 1.4 MB of data. UNIVACs had a clock speed of 2.2 MHz and a memory (mercury acoustic delay lines) of 1000 x 12 digit words (i.e., ~12 KB). It could complete 8,333 additions of approximately 100 bit words (11 decimal digits plus sign) in 1 second (8,300 Hz or 8.3 KHz). The smaller but newer Burroughs machine I learned to program on in 1958-59 had about the same power. Today (January 2002), I am writing this document on a $1,000 Pentium 4 computer that has more than 3.8 billion times more raw processing power than was available to one of the nation's largest life insurance companies 50 years ago from a million dollar room full of electronics81! And yet, it was apparently cost-effective for the insurance company to make that investment.

Manchester University and Ferranti in the UK also have a credible claim to have produced the first commercial computer in 195182, based on development work in the Electrical Engineering and Mathematics Departments of Manchester University83. Some of the key people who were involved in the German code breaking efforts at Bletchley Park joined the work at Manchester University. The Manchester/Ferranti Mark I design was based on the use of cathode ray tubes (CRT) for "core" memory and magnetic drum storage for working files. Alan Turing, one of the greatest mathematicians of the 20th Century, also heavily involved in the German codebreaking effort at Blechley Park, helped develop the coding system for the Mark I and wrote the first programmers manual for it84.

The concept of computer (hardware) generations is a useful way of summarising the multiple revolutions in technology that followed85.

The pace of technological innovation in creating the first generation of computers of the1950's was rapid (Lee et al. 1995). IBM introduced its first electronic computer, the 701, in 195286. Like the Manchester/Ferranti Mark 1, the IBM 701 used elecrostatic CRT's, magnetic drums and/or magnetic cores for memory; and pioneered magnetic tape drives for bulk storage. The drives used vacuum columns to draw the plastic-backed tape out in a long loop between the reels to minimise tape breakage from the frequent back and forth stop and start operation of the tapes87. Although processing power was limited, the tape drives allowed the computers to efficiently process and manage large volumes of records by comparison to the mechanical sorting and tabulating of punch cards. However, punch cards still provided the primary tool for entering information into the computing environment.

My first experience with electronic computers was in 1958-59 on a first-generation machine in the Burroughs 204-205 series, first marketed in 1954. Programs and working data were all held within a 1000 word x 32 bit magnetic drum memory (fully configured 204-205 machines had 4000 word by 40 bit drums). However, the machines were also equipped with vacuum column magnetic tape drives. The human input was via an electric typewriter keyboard to paper tape punch, which was then written to magnetic or memory via a paper tape reader. All of the first generation computers were programmed directly using machine code.

Second Generation Computers (1955 - 1964)