digital computer

n.
A computer that performs calculations and logical operations with quantities represented as digits, usually in the binary number system.

For more information on digital computer, visit Britannica.com.

A device that processes numerical information; more generally, any device that manipulates symbolic information according to specified computational procedures. The term digital computer—or simply, computer—embraces calculators, computer workstations, control computers (controllers) for applications such as domestic appliances and industrial processes, data-processing systems, microcomputers, microcontrollers, multiprocessors, parallel computers, personal computers, network servers, and supercomputers. See also Calculators; Digital control; Microcomputer; Programmable controllers; Supercomputer.

A digital computer is an electronic computing machine that uses the binary digits (bits) 0 and 1 to represent all forms of information internally in digital form. Every computer has a set of instructions that define the basic functions it can perform. Sequences of these instructions constitute machine-language programs that can be stored in the computer and used to tailor it to an essentially unlimited number of specialized applications. Calculators are small computers specialized for mathematical computations. General-purpose computers range from pocket-sized personal digital assistants (notepad computers), to medium-sized desktop computers (personal computers and workstations), to large, powerful computers that are shared by many users via a computer network. The vast majority of digital computers now in use are inexpensive, special-purpose microcontrollers that are embedded, often invisibly, in such devices as toys, consumer electronic equipment, and automobiles. See also Bit; Computer programming; Embedded systems.

The main data-processing elements of a computer reside in a small number of electronic integrated circuits (ICs) that form a microprocessor or central processing unit (CPU). Electronic technology allows a basic instruction such as “add two numbers” to be executed many millions of times per second. Other electronic devices are used for program and data storage (memory circuits) and for communication with external devices and human users (input-output circuits). Nonelectronic (magnetic, optical, and mechanical) devices also appear in computers. They are used to construct input-output devices such as keyboards, monitors (video screens), secondary memories, printers, sensors, and mechanical actuators.

Information is stored and processed by computers in fixed-sized units called words. Common word sizes are 8, 16, 32, and 64 bits. Four-bit words can be used to encode the first 16 integers. By increasing the word size, the number of different items that can be represented and their precision can be made as large as desired. A common word size in personal computers is 32 bits, which allows 2³² = 4,294,967,296 distinct numbers to be represented.

Computer words can represent many different forms of information, not just numbers. For example, 8-bit words called characters or bytes are used to encode text symbols (the 10 decimal digits, the 52 upper- and lowercase letters of the English alphabet, and punctuation marks). A widely used code of this type is ASCII (American Standard Code for Information Interchange). Visual information can be reduced to black and white dots (pixels) corresponding to 0's and 1's. Audio information can be digitized by mapping a small element of sound into a binary word; for example, a compact disk (CD) uses several million 16-bit words to store an audio recording. Logical quantities encountered in reasoning or decision making can be captured by associating 1 with true and 0 with false. Hence, most forms of information are readily reduced to a common, numberlike binary format suitable for processing by computer. See also Compact disk; Image processing; Logic.

Logic components

The operation of a digital computer can be viewed at various levels of abstraction, which are characterized by components of different complexity. These levels range from the low, transistor level seen by an electronic circuit designer to the high, system level seen by a computer user. A useful intermediate level is the logic level, where the basic components process individual bits. By using other basic components called gates, logic circuits can be constructed to perform many useful operations. See also Logic circuits.

System organization

An accumulator is a digital system that constitutes a simple processor capable of executing a few instructions. By introducing more data-processing circuits and registers, as well as control circuits for a larger set of instructions, a practical, general-purpose processor can be constructed. Such a processor forms the “brain” of every computer, and is referred to as its central processing unit. A CPU implemented on a single integrated-circuit chip is called a microprocessor. See also Microprocessor.

A typical computer program is too large to store in the CPU, so another component called the main memory is used to store a program's instructions and associated data while they are being executed (Fig. 1). Main memory consists of high-speed integrated circuits designed to allow storage and retrieval of information one word at a time. All words in main memory can be accessed with equal ease; hence this is also called a random-access memory (RAM).

General organization of a computer.

A computer program is processed by loading it into main memory and then transferring its instructions and data one word (or a few words) at a time to the CPU for processing. Hence, there is a continual flow of instructions and data words between the CPU and its main memory. As millions of words must be transferred per second, a high-speed communication link is needed between the CPU and main memory. The system bus (Fig. 1) fills this role.

A computer has input-output (I/O) control circuits and buses to connect it to external input-output devices (also called peripherals). Typical input-output devices are a keyboard, which is an input device, and a printer, which is an output device. Because most computers need more storage space than main memory can supply, they also employ secondary memory units which form part of the computer's input-output subsystem. Common secondary memory devices are hard disk drives, flexible (floppy) disk drives, and magnetic tape units. Compared to main memory, secondary memories employ storage media (magnetic disks and tapes) that have higher capacity and lower cost. However, secondary memories are also significantly slower than main memory. See also Computer peripheral devices; Computer storage technology.

No explicit instructions are needed for input-output operations if input-output devices share with main memory the available memory addresses. This is known as memory-mapped input-output, and allows load and store instructions to be used to transfer data between the CPU and input-output devices. In general, a computer's instruction set should include a selection of instructions of the following three types: (1) Data-transfer instructions that move data unchanged between the CPU, main memory, and input-output devices. (2) Data-processing instructions that perform numerical operations such as add, subtract, multiply, and divide, as well as nonnumerical (logical) operations, such as NOT, AND, EXCLUSIVE-OR, and SHIFT. (3) Program-control instructions that can change the order in which instructions are executed, for example branch, branch-on-zero, call procedure, and return from procedure.

The instruction unit (I unit) of a CPU (Fig. 2), also called the program control unit, is responsible for fetching instructions from main memory, using the program counter as the instruction address register. The opcode of a newly fetched instruction I is placed in the instruction register. The opcode is then decoded to determine the sequence of actions required to execute I. These may include the loading or storing of data assigned to main memory, in which case the I unit computes all needed addresses and issues all needed control signals to the CPU and the system bus. Data are processed in the CPU's execution unit (E unit), also called the datapath, which contains a set of registers used for temporary storage of data operands, and an arithmetic logic unit (ALU), which contains the main data-processing circuits.

Internal organization of a CPU.

Performance measures

A simple indicator of a CPU's performance is the frequency f of its central timing signal (clock), measured in millions of clock signals issued per second or megahertz (MHz). The clock frequency depends on the integrated-circuit technology used; frequencies of several hundred megahertz are achievable with current technology. Each clock signal triggers execution of a basic instruction such as a fixed-point addition; hence, the time required to execute such an instruction (the clock cycle time) is 1/f microseconds. Complex instructions like multiplication or operations on floating-point numbers require several clock cycles to complete their execution. Another measure of CPU performance is the (average) instruction execution rate, measured in millions of instructions per second (MIPS).

Instruction execution time is strongly affected by the time to move instructions or data between the CPU and main memory. The time required by the CPU to access a word in main memory is typically about five times longer than the CPU's clock cycle time. This disparity in speed has existed since the earliest computers despite efforts to develop memory circuits that would be fast enough to keep up with the fastest CPUs. Maximum performance requires the CPU to be supplied with a steady flow of instructions that need to be executed. This flow is disrupted by branch instructions, which account for 20% or more of the instructions in a typical program.

To deal with the foregoing issues, various performance-enhancing features have been incorporated into the design of computers. The communication bottleneck between the CPU and main memory is reduced by means of a cache, which is a special memory unit inserted between the two units. The cache is smaller than main memory but can be accessed more rapidly, and is often placed on the same integrated-circuit chip as the CPU. Its effect is to reduce the average time required by the CPU to send information to or receive information from the memory subsystem. Special logic circuits support the complex flow of information among main memory, the cache, and the registers of the CPU. However, the cache is largely invisible to the programs being executed.

The instruction execution rate can be increased by executing several instructions concurrently. One approach is to employ several E units that are tailored to different instruction types. Examples are an integer unit designed to execute fixed-point instructions and a floating-point unit designed for floating-point instructions. The CPU can then execute a fixed-point instruction and a floating-point instruction at the same time. Processors that execute several instructions in parallel in this way are called superscalar. See also Concurrent processing.

Another speedup technique called pipelining allows several instructions to be processed simultaneously in special circuits called pipelines. Execution of an instruction is broken into several consecutive steps, each of which can be assigned to a separate stage of the pipeline. This makes it possible for an n-stage E unit to overlap the execution of up to n different instructions. A pipeline processing circuit resembles an assembly line on which many products are in various stages of manufacture at the same time. The ability of a CPU to execute several instructions at the same time by using multiple or pipelined E units is highly dependent on the availability of instructions of the right type at the right time in the program being executed. A useful measure of the performance of a CPU that employs internal parallelism is the average number of clock cycles per instruction (CPI) needed to execute a representative set of programs.

CISCs and RISCs

A software implementation of a complex operation like multiply is slower than the corresponding hardware implementation. Consequently, as advances in IC technology lowered the cost of hardware circuits, instruction sets tended to increase in size and complexity. By the mid-1980s, many microprocessors had instructions of several hundred different types, characterized by diverse formats, memory addressing modes, and execution times. The heterogeneous instruction sets of these complex instruction set computers (CISCs) have some disadvantages. Complex instructions require more processing circuits, which tend to make CISCs large and expensive. Moreover, the decoding and execution of complex instruction can slow down the processing of simple instructions.

To address the defects of CISCs, a new class of fast computers referred to as reduced instruction set computers (RISCs) was introduced. RISCs are characterized by fast, efficient—but not necessarily small—instruction sets. The following features are common to most RISCs: (1) All instructions are of fixed length and have just a few opcode formats and addressing modes. (2) The only instructions that address memory are load and store instructions; all other instructions require their operands to be placed in CPU registers. (3) The fetching and processing of most instructions is overlapped in pipelined fashion.

---

digital computer

A computer that performs operations with quantities represented electronically as digits, usually in binary code. (Compare analog computer.)

A computer that accepts and processes data that has been converted into binary numbers. Most computers are digital. Contrast with analog computer.

Download Computer Desktop Encyclopedia to your iPhone/iTouch

Type of Computer that represents information in discrete form, as opposed to an analog computer, which allows representations to vary along a continuum. All modern general-purpose electronic computers are digital.

One of the two types of computers, the other being an analog computer. A digital computer operates by counting data represented by coded characters such as numbers, letters, and symbols.

n

A computer that operates on discrete data by performing arithmetic and logic processes on them.