Early modern computers are typically grouped into four "generations." Each generation is marked by improvements in basic technology. These improvements in technology have been extraordinary and each advance has resulted in computers of lower cost, higher speed, greater memory capacity, and smaller size.
This grouping into generations is not clear-cut nor is it without debate. Many of the inventions and discoveries that contributed to the modern computer era do not neatly fit into these strict categories. The reader should not interpret these dates as strict historical boundaries.
First Generation (1945–1959)
The vacuum tube was invented in 1906 by an electrical engineer named Lee De Forest (1873–1961). During the first half of the twentieth century, it was the fundamental technology that was used to construct radios, televisions, radar, X-ray machines, and a wide variety of other electronic devices. It is also the primary technology associated with the first generation of computing machines.
The first operational electronic general-purpose computer, named the ENIAC (Electronic Numerical Integrator and Computer), was built in 1943 and used 18,000 vacuum tubes. It was constructed with government funding at the University of Pennsylvania's Moore School of Engineering, and its chief designers were J. Presper Eckert, Jr. (1919–1995) and John W. Mauchly (1907–1980). It was almost 30.5 meters (100 feet) long and had twenty 10-digit registers for temporary calculations. It used punched cards for input and output and was programmed with plug board wiring. The ENIAC was able to compute at the rate of 1,900 additions per second. It was used primarily for war-related computations such as the construction of ballistic firing tables and calculations to aid in the building of the atomic bomb.
The Colossus was another machine that was built during these years to help fight World War II. A British machine, it was used to help decode secret enemy messages. Using 1,500 vacuum tubes, the machine, like the ENIAC, was programmed using plug board wiring.
These early machines were typically controlled by plug board wiring or by a series of directions encoded on paper tape. Certain computations would require one wiring while other computations would require another. So, while these machines were clearly programmable, their programs were not stored internally. This would change with the development of the stored program computer.
The team working on the ENIAC was probably the first to recognize the importance of the stored program concept. Some of the people involved in the early developments of this concept were J. Presper Eckert Jr. (1919–1955) and John W. Mauchly (1907–1980), and John von Neumann (1903–1957). During the summer of 1946, a seminar was held at the Moore School that focused great attention on the design of a stored program computer. About thirty scientists from both sides of the Atlantic Ocean attended these discussions and several stored programmed machines were soon built.
One of the attendees at the Moore School seminar, Maurice Wilkes (1913–), led a British team that built the EDSAC (Electronic Delay Storage Automatic Calculator) at Cambridge in 1949. On the American side, Richard Snyder led the team that completed the EDVAC (Electronic Discrete Variable Automatic Computer) at the Moore School. Von Neumann helped design the IAS (Institute for Advanced Study) machine that was built at Princeton University in 1952. These machines, while still using vacuum tubes, were all built so that their programs could be stored internally.
Another important stored program machine of this generation was the UNIVAC (UNIVersal Automatic Computer). It was the first successful commercially available machine. The UNIVAC was designed by Eckert and Mauchly. It used more than 5,000 vacuum tubes and employed magnetic tape for bulk storage. The machine was used for tasks such as accounting, actuarial table computation, and election prediction. Forty-six of these machines were eventually installed.
The UNIVAC, which ran its first program in 1949, was able to execute ten times as many additions per second as the ENIAC. In modern dollars, the UNIVAC was priced at $4,996,000. Also, during this period, the first IBM computer was shipped. It was called the IBM 701 and nineteen of these machines were sold.
Second Generation (1960–1964)
As commercial interest in computer technology intensified during the late 1950s and 1960s, the second generation of computer technology was introduced—based not on vacuum tubes but on transistors .
John Bardeen (1908–1991), William B. Shockley (1910–1989), and Walter H. Brattain (1902–1987) invented the transistor at Bell Telephone Laboratories in the mid-1940s. By 1948 it was obvious to many that the transistor would probably replace the vacuum tube in devices such as radios, television sets, and computers.
One of the first computing machines based on the transistor was the Philco Corporation's Transac S-2000 in 1958. IBM soon followed with the transistor-based IBM 7090. These second generation machines were programmed in languages such as COBOL (Common Business Oriented Language) and FORTRAN (Formula Translator) and were used for a wide variety of business and scientific tasks. Magnetic disks and tape were often used for data storage.
Third Generation (1964–1970)
The third generation of computer technology was based on integrated circuit technology and extended from approximately 1964 to 1970. Jack Kilby (1923–) of Texas Instruments and Robert Noyce (1927–1990) of Fairchild Semiconductor were the first to develop the idea of the integrated circuit in 1959. The integrated circuit is a single device that contains many transistors.
Arguably the most important machine built during this period was the IBM System/360. Some say that this machine single handedly introduced the third generation. It was not simply a new computer but a new approach to computer design. It introduced a single computer architecture over a range or family of devices. In other words, a program designed to run on one machine in the family could also run on all of the others. IBM spent approximately $5 billion to develop the System/360.
One member of the family, the IBM System/360 Model 50, was able to execute 500,000 additions per second at a price in today's dollars of $4,140,257. This computer was about 263 times as fast as the ENIAC.
During the third generation of computers, the central processor was constructed by using many integrated circuits. It was not until the fourth generation that an entire processor would be placed on a single silicon chip—smaller than a postage stamp.
Fourth Generation (1970–?)
The fourth generation of computer technology is based on the microprocessor. Microprocessors employ Large Scale Integration (LSI) and Very Large Scale Integration (VLSI) techniques to pack thousands or millions of transistors on a single chip.
The Intel 4004 was the first processor to be built on a single silicon chip. It contained 2,300 transistors. Built in 1971, it marked the beginning of a generation of computers whose lineage would stretch to the current day.
In 1981 IBM selected the Intel Corporation as the builder of the microprocessor (the Intel 8086) for its new machine, the IBM-PC. This new computer was able to execute 240,000 additions per second. Although much slower than the computers in the IBM 360 family, this computer cost only $4,000 in today's dollars! This price/performance ratio caused a boom in the personal computer market.
In 1996, the Intel Corporation's Pentium Pro PC was able to execute 400,000,000 additions per second. This was about 210,000 times as fast as the ENIAC–the workhorse of World War II. The machine cost only $4,400 in inflation-adjusted dollars.
Microprocessor technology is now found in all modern computers. The chips themselves can be made inexpensively and in large quantities. Processor chips are used as central processors and memory chips are used for dynamic random access memory (RAM) . Both types of chips make use of the millions of transistors etched on their silicon surface. The future could bring chips that combine the processor and the memory on a single silicon die.
During the late 1980s and into the 1990s cached, pipelined, and superscaler microprocessors became commonplace. Because many transistors could be concentrated in a very small space, scientists were able to design these single chip processors with on-board memory (called a cache ) and were able to exploit instruction level parallelism by using instruction pipelines along with designs that permitted more than one instruction to be executed at a time (called superscaler). The Intel Pentium Pro PC was a cached, superscaler, pipelined microprocessor.
Also, during this period, an increase in the use of parallel processors has occurred. These machines combine many processors, linked in various ways, to compute results in parallel. They have been used for scientific computations and are now being used for database and file servers as well. They are not as ubiquitous as uniprocessors because, after many years of research, they are still very hard to program and many problems may not lend themselves to a parallel solution.
The early developments in computer technology were based on revolutionary advances in technology. Inventions and new technology were the driving force. The more recent developments are probably best viewed as evolutionary rather than revolutionary.
It has been suggested that if the airline industry had improved at the same rate as the computer industry, one could travel from New York to San Franscisco in 5 seconds for 50 cents. In the late 1990s, microprocessors were improving in performance at the rate of 55 percent per year. If that trend continues, and it is not absolutely certain that it will, by the year 2020 a single microprocessor could possess all the computing power of all the computers in Silicon Valley at the dawn of the twenty-first century.
see also Apple Computer, Inc.; Bell Labs; Eckert, J. Presper, Jr. and Mauchly, John W.; Integrated Circuits; Intel Corporation; Microsoft Corporation; Xerox Corporation.
Michael J. McCarthy
Hennessy, John, and David Patterson. Computer Organization and Design. San Francisco: Morgan Kaufmann Publishers, 1998.
Rockett, Frank H. "The Transistor." Scientific American 179, no. 3 (1948): 52.
Williams, Michael R. A History of Computing Technology. Los Alamitos, CA: IEEE Computer Society Press, 1997.