Encyclopedia of Computer Science and Technology

history of computing 227electromechanical Mark I. ENIAC had about 19,000 vacuumtubes and consumed as much power as perhaps a thousandmodern desktop PCs.1950sThe 1950s saw the establishment of a small but viable commercialcomputer industry in the United States and parts ofEurope. Eckert and Mauchly formed a company to designand market the UNIVAC, based partly on work on the experimentalEDVAC. This new generation of computers wouldincorporate the key concept of the stored program: Ratherthan the program being set up by wiring or simply readsequentially from tape or cards, the program instructionswould be stored in memory just like any other data. Besidesallowing a computer to fetch instructions at electronic ratherthan mechanical speeds, storing programs in memory meantthat one part of a program could refer to another part duringoperation, allowing for such mechanisms as branching,looping, the running of subroutines, and even the ability ofa program to modify its own instructions.The UNIVAC became a hit with the public when it wasused to correctly predict the outcome of the 1952 presidentialelection. Government offices and large corporationsbegan to look toward the computer as a way to solve theirincreasingly complex data processing needs. Forty UNI-VACs were eventually built and sold to such customers asthe U.S. Census Bureau, the U.S. Army and Air Force, andinsurance companies. Sperry (having bought the Mauchly-Eckert company), Bendix, and other companies had somesuccess in selling computers (often for specialized applications),but it was IBM that eventually captured the broadbusiness market for mainframe computers.The IBM 701 (marketed to the government and defenseindustry) and 702 (for the business market) incorporated severalemerging technologies including a fast electronic (tube)memory that could store 4,096 36-bit data words, a rotatingmagnetic drum that could store data that is not immediatelyneeded, and magnetic tape for backup. The IBM 650, marketedstarting in 1954, became the (relatively) inexpensiveworkhorse computer for businesses (see mainframe). TheIBM 704, introduced in 1955, incorporated magnetic corememory and also featured floating-point calculations.1960sThe 1960s saw the advent of a “solid state” computer designfeaturing transistors in place of vacuum tubes and the useof ferrite magnetic core memory (introduced commerciallyin 1955). These innovations made computers both morecompact (although they were still large by modern standards),more reliable, and less expensive to operate (dueto lower power consumption.) The IBM 1401 was a typicalexample of this new technology: It was compact, relativelysimple to operate, and came with a fast printer that made iteasier to generate data.There was a natural tendency to increase the capacityof computers by adding more transistors, but the handwiringof thousands of individual transistors was difficultand expensive. As the decade progressed, however, the conceptof the integrated circuit began to be implemented incomputing. The first step in that direction was to attach anumber of transistors and other components to a ceramicsubstrate, creating modules that could be handled andwired more easily during the assembly process.IBM applied this technology to create what wouldbecome one of the most versatile and successful lines inthe history of computing, the IBM System/360 computer.This was actually a series of 14 models that offered successivelygreater memory capacity and processing speedwhile maintaining compatibility so that programs developedon a smaller, cheaper model would also run on themore expensive machines. Compatibility was ensured bydevising a single 360 instruction set that was implementedat the machine level by microcode stored in ROM (read-onlymemory) and optimized for each model. By 1970 IBM hadsold more than 18,000 360 systems worldwide.By the mid-1960s, however, a new market segment hadcome into being: the minicomputer. Pioneered by DigitalEquipment Corporation (DEC) with its PDP line, the minicomputerwas made possible by rugged, compact solid-state(and increasingly integrated) circuits. Architecturally, themini usually had a shorter data word length than the mainframe,and used indirect addressing (see addressing) forflexibility in accessing memory. Minis were practical for usesin offices and research labs that could not afford (or house)a mainframe (see minicomputer). They were also a boonto the emerging use of computers in automating manufacturing,data collection, and other activities, because a minicould fit into a rack with other equipment (see also embeddedsystems). In addition to DEC, Control Data Corporation(CDC) produced both minis and large high-performancemachines (the Cyber series), the first truly commerciallyviable supercomputers (see supercomputer).In programming, the main innovation of the 1960s wasthe promulgation of the first widely-used, high-level programminglanguages, COBOL (for business) and FORTRAN(for scientific and engineering calculations), the result ofresearch in the late 1950s. While some progress had beenmade earlier in the decade in using symbolic names forquantities and memory locations (see assembler), the newhigher-level languages made it easier for professionals outsidethe computer field to learn to program and made theprograms themselves more readable, and thus easier tomaintain. The invention of the compiler (a program thatcould read other programs and translate them into lowlevelmachine instructions) was yet another fruit of thestored program concept.1970sThe 1970s saw minis becoming more powerful and versatile.The DEC VAX (“Virtual Address Extension”) series allowedlarger amounts of memory to be addressed and increasedflexibility. Meanwhile, at the high end, Seymour Cray leftCDC to form Cray Research, a company that would producethe world’s fastest supercomputer, the compact, freoncooledCray-1. In the mainframe mainstream, IBM’s 370series maintained that company’s dominant market share inbusiness computing.