Automatic Electronic Computers ESSENTIAL SIMPLICITY KEY TO INSTRUMENT’S POWER

Press, Volume CIII, Issue 30535, 2 September 1964, Page 16

Automatic Electronic Computers ESSENTIAL SIMPLICITY KEY TO INSTRUMENT’S POWER

(Specialty written for "The Press" bv B. A M M°°N. Senior Lecturer in Mathematic,. University of Canterbury.] Everyone agrees that automatic electronic computers are here to stay; most people have heard something of the remarkable range of tasks they can perform; many people have actually seen one at work. But there are still a lot of people who would like to have a straight-forward answer to the question: what actually is a computer? Some peopte are Reefton schoolchildren, on whose behalf a reader of The Press asked this question. This article tries to answer it.

If one sets out to find the familiar object whose, properties display most closely the essential nature of the automatic digital computer, one finds that this is not the desk calculator, as might have been expected, but the automatic telephone exchange. The mere ability to calculate is almost incidental. The telephone exchange is a control unit containing a highly systematic pattern of circuits which can be set up In a particular way to perform one of a large number of possible operations (e.g., the ringing of a bell on a particular receiver), when a certain coded set- of numbers is entered into it. Electrical Pulses A control unit is also an essential part of a computer, and, indeed, some of the very earliest computers were constructed of the same electro-mechanical units as a telephone exchange. The coded sets of numbers entered into a computer’s control unit are called instructions.

It was, however, the tremendous development in electronics, especially radar, in the last war that turned the computer into a practical concept. In the circuits of a computer control unit the numerically-coded instructions can be represented by sequences of electrical pulses. Such pulses, by making certain valves or transistors conductive, can open or close electrical circuits. In this way succeeding pulses can be guided through the system, to perform in turn their own control functions."

Numbers can be represented in pulse patterns, by the presence or absence of pulses at particular points in the pulse sequence. The presence or absence of a pulse corresponds to two distinct numerals, one or zero, Which are all that are required for binary, or base two, numbers. Such numbers are thus the natural and most efficient form of numeration for a computer. Switching Network The computer control unit is in fact a large switching network which in some respects can be compared with a giant railway shunting yard. The essential feature of such a switching network is its logical nature. If a pulse is present, a circuit is made to conduct; if not, the circuit does not conduct To take the

railway analogy: if a goods train approaches, switch to a side line; if a passenger train, switch to the main line. If there is a green light, proceed; if red, stop. In numerical work the presence of a certain pulse may represent a positive number and its absence a negative number. Many logical decisions depend on whether a number is positive or negative, for example, a credit or a debit balance. The corresponding alternative courses of action in a computer can be governed by the presence or absence of an electrical pulse. ~ , We have now the key which tells us precisely what a computer can and cannot do. If an operation, no matter how complex, can be described by a finite series of simple logical steps from a known initial state, a computer can carry it out; otherwise it cannot. In principle this includes all the operations of banking and commerce, stock control and payrolls, scheduling of construction projects, traffic control by road, rail, and air, the calculations of scientific research and engineering design, deciphering of codes, control of oil refineries. There is almost no end to this list.

Clearly, the operations of arithmetic are logical and thus can be carried out by a computer. In binary arithmetic these logical rules are especially simple. Thus, if either of two digits to be added is -zero, the sum is equal to the other one; otherwise the answer is zero and carry one. In multiplication, if either digit is zero, the answer is zero, otherwise it is one. The “Building Bricks” A programme for a computer is a sequence of numer-ically-coded instructions each of which specifies an elementary logical or arithmetic instruction or a transfer of data. Instructions are the building-bricks of intricately complex computer programmes just as words are the bricks of which sonnets and sagas, novels and narratives are made. The point at which an electronic computer differs markedly from a telephone exchange is in its speed. It takes several seconds to dial a single instruction into a telephone exchange. In the same time a computer can execute many thousand different instructions. Clearly, therefore, all its instructions must be made available to .it at a comparable speed; that is to say, this process must be automatic. It is for this reason that a computer possesses a number-storage or memory unit, into which all the num-erically-coded instructions for a programme must be loaded before that programme can be executed.

Nowadays such numberstorage units consist of a large number of magnetic cores: small rings composed of magnetic oxide of iron, the lodestone of the ancient mariner, strung on current-carrying wires. A pulse of electricity flowing one way will magnetize a core clockwise, and the other way, anticlockwise —ideal to represent the numerals zero and one re-

quired in the binary number system. Arithmetic Unit Since one of the functions of the computer is to do arithmetic, it must also have a unit in which arithmetic operations upon numbers may be carried out. This is its arithmetic unit, the equivalent of a desk calculator—the third functional element of a computer. The numbers operated upon are known as the data. Since the arithmetic operations are carried out at very high speed, the data must also be available automatically; so the data, being numbers like the instructions, are also placed in storage. What distinguishes data and instructions in storage is not the way they are represented, for that is indistinguishable. It is the way they are treated. Instructions are routed in sequence to the control unit to specify the operations to be carried out. Data, under control of the instructions, are routed to the arithmetic unit to be added, compared, multiplied, and so on, the answer being transferred back into storage. One of the exciting things about computers is that the numbers in storage representing instructions can, under the control of other instructions, be operated upon arithmetically just like data. In this way the operations they represent are modified—that is, the machine can modify its course of action as it proceeds. This is the basis of the phenomenon known as “machine learning.” We do not yet know the ultimate potential of this property. Finally, to make it complete, the computer needs an input unit through which the instructions and input data may be entered into storage, and an output unit at which the output data or answers are delivered. Such units to read or punch cards or paper tape, read magnetic tape, or print documents are becoming increasingly familiar. The range of input and output media continues to widen; even successful voice communication has been achieved. A computer may have several different input and output units. Communication Method It remains to consider how, if the idiomatic language of computers consists of no more than sequences of ones and zeros, it is possible for human beings to communicate freely with them. The answer is that if we can state formally in English and algebra-like Statements the sequence of operations we wish to have carried out, then it is possible to devise a computer programme to carry out the logical procedure of translating these statements to instructions in its own numerical code. Thus the computer itself plays a major role in the task of writing its own programmes. The key to the power and success of the modern automatic computer is its essential simplicity and generality. One of the things most needed in this world today is the education and the wisdom to permit us to use this great power wisely and well.