Encyclopedia of Computer Science and Technology

304 microprocessorMicroprocessor development began in the 1960s whena new company called Intel was given a contract to developchips for programmable calculators for a Japanese firm.Marcian E. “Ted” Hoff headed the project. He decided thatrather than hard-wiring most of the calculator logic into thechips, he would create a general-purpose chip that couldread instructions and data, perform basic arithmetic andlogical functions, and transfer data between memory andinternal locations called registers.The resulting microprocessor, when combined with someRAM (random access memory), some preprogrammed ROM(Read Only Memory), and an input/output (I/O) chip constituteda tiny but complete CPU, soon dubbed “a computeron a chip.” This first microprocessor, the Intel 4004, hadonly a few thousand transistors, could handle data only 4bits at a time, and ran at 740 KHz (about one three-thousandththe speed of the latest Pentium IV chips).Intel gradually refined the chip, giving it the logic circuitsto enable it to perform additional instructions, moreinternal stack and register space, and 8 KB of space to storeprograms. The 8008 could handle 8 bits of data at a time,while the 8080 became the first microprocessor that wascapable of serving as the CPU for a practical microcomputersystem. Its descendants, the 8088 and 8086 (16-bit) poweredindustry-standard IBM-compatible PCs. Meanwhile,other companies such as Motorola (68000), Zilog (Z-80),and MOS Technology (6502) powered competing PCs fromApple, Atari, Commodore, and others.With the dominance of the IBM PC and its clones (seeibm pc), the Intel 80 × 86 series in turn dominated themicroprocessor market. (The x refers to successive digits, asin 80286, 80386, and 80486.) At the next level this nomenclaturewas replaced by the Pentium series, which is up tothe Pentium 4 as of 2002.The MITS Altair (1975) was the first microcomputer availablecommercially. It was generally purchased in kit form. While theAltair did not have much processing capacity, it aroused great interestand inspired other computer builders such as Apple’s Steve Wozniakand Steve Jobs. (Christopher Fitzgerald / The ImageWorks)According to a famous dictum called Moore’s Law, thedensity (number of transistors per cubic area) and speed (interms of clock rate) of microprocessors has roughly doubledevery 18 months to two years. Intel expects to be makingmicroprocessors with 1 billion transistors by 2007.Microprocessor and MicrocomputerA microcomputer is a system consisting of a microprocessorand a number of auxiliary chips. The microprocessorchip serves as the central processing unit (CPU). It containsa clock that regulates the flow of data and instructions(each instruction takes a certain number of clock cycles toexecute). There is also an index register that keeps track ofthe instruction being executed. A small number of locationscalled registers within the CPU allow for storing or retrievingthe data being used by instructions much more quicklythan retrieval from main memory (RAM).Typically, the instruction register advances to the nextinstruction. The instruction is fetched, decoded, and sent tothe CPU’s ALU (arithmetic logic unit) for processing. Dataneeded to be processed by the instruction are either fetchedfrom a register or, through an address register, fetched fromRAM. (Some processors store one operand for an arithmeticoperation in a special register called the accumulator.)Floating-point operations (those involving numbersthat can include decimal points) require special registersthat can keep track of the decimal position. Until the mid-1990s, many systems used a separate microprocessor calleda coprocessor to handle floating point operations. However,later chips such as the Pentium series integrate floatingpoint operations into the main chip.In order to function as the heart of a microcomputer,the CPU must communicate with a variety of other devicesby interacting with special controller chips. For example,there is a bus interface chip (see bus) that decodes memoryaddresses and routes requests to the appropriate devices onthe motherboard. When data is requested from memory, amemory controller must physically fetch the data from RAM(see memory). There is also a cache controller that interfaceswith one or two levels of high-speed cache memory (seecache). The algorithms implemented in the cache controlleraim to have the next instructions and the most-likely neededdata already in the cache when the CPU requests them.Other devices such as disk drives, modems, printers,and video cards are all connected to the CPU through input/output (I/O) interfaces that connect to the system bus. Mostof the devices connected to the bus have their own microprocessors.Software (see device driver) translates highlevelprogramming instructions (such as to open a file) tothe appropriate device commands.The CPU and many other devices also contain ROM (readonly memory) chips that have permanent basic instructionsstored on them (see bios). This enables the CPU and otherdevices to perform the necessary actions to enter into communicationwhen the system starts up (see boot sequence).New Features EmergeImprovements in microprocessors during the 1980sincluded wider data paths and the ability to address a larger