Home |
By ROB KATZ, Applications Engineer. Digital Equipment Corp. It can be said that a minicomputer has the basic advantages of a desk-top calculator and time-sharing system but has remarkably few of the disadvantages of either. Editor's Note: This is the second article on the subject of the computer best suited to the engineer. Last month, Paul Asmus of Hewlett-Packard covered the "programmable calculator. " Next month Lee N. Beyer of General Electric will discuss the pros and cons of time-sharing systems.
THE image of the electronic computer is changing. Along with the huge, monolithic pieces of machinery usually found in computation centers (and wildly parodied in movies), a new breed of machine has made its appearance. This newcomer is considerably smaller (sometimes small enough to fit on a desk top), is usually more rugged, and has been adapted to a number of unusual uses. Variously called "small general-purpose computers," "little computers," and "baby computers" by some manufacturers and users, these devices are almost universally dubbed "minicomputers" by the electronics industry. However, since they come in various shapes and sizes, there is some confusion as to just what constitutes a mini-computer. A minicomputer is a complete computer, capable of performing normal computer functions independently (as opposed to, say, an exotic terminal that has to be connected to a large computer in order to work); and it is general purpose, meaning that whatever functions it may do are determined by the program that is read into it-and that its functions can change merely by reading another program into it. By contrast, a special-purpose computer, whose functions cannot be reprogrammed directly, is not considered a mini-computer. Another aspect that identifies a minicomputer is its word length. Although all digital computers use binary information units (bits) for their computation these are grouped into larger units of bits called "words." The typical word length of a large computer is 32-36 bits, the minicomputer from 8 to 18 bits. The minicomputer memory has a capacity of at least 4096 words (4K words in computer parlance). A final parameter defining a minicomputer is that it generally costs less than $25,000 in its basic configuration. Admittedly, this last parameter is somewhat arbitrary, but it draws a line in the rather extensive gray area where a minicomputer leaves off and the larger "medium-scale" or "midi" computer starts. It is important to emphasize "basic configuration" since quite frequently the minicomputer uses supplemental devices to help it do its job-known in the industry as "peripheral devices" or "peripherals." This POP-8/E can be used as an independent terminal for engineering calculations in this configuration. With the modern programming languages available, such as Focal, the computer can be programmed by engineer for use by untrained personnel. Peripherals are used to store data or programs or to input and output information to the computer. For storage, the basic method is magnetic recording. Magnetic tapes, disks, and drums are all used to record and play back information from and to the computer. With such storage media, it is possible to keep available for rapid recall program elements or data that may be used only from time to time, thus keeping the computer's internal memory free for more immediate program needs. In order of recording and playback data rates, drums are the fastest, followed by disks and tape in that order. Tapes and some disks can be used to transfer data from one computer to another in the same way that it is possible to play back an audio tape recording on a machine other than the one on which it was recorded. Thus, in addition to being storage devices, some disks and tapes are used for data transfer. Sometimes when a small computer system is logging data from a process, it will prepare tapes so larger computers located elsewhere can make a more detailed analysis at a later time. The fundamental peripheral device by which one communicates with the computer is a Teletype or its equivalent. Although almost all minicomputers come equipped with some such device, it is considered a peripheral because it is tied to the computer in the same way as all other peripherals-that is, through appropriate interfacing. Besides the basic teletypewriter, other more specialized and sophisticated devices are used. High-speed line printers are used to output large quantities of printed data in a short time. Graphical plotters are used to output information in graphic form, which can be in Cartesian or polar coordinates-or even isometrics or perspective drawings. All these peripherals put out printed or "hardcopy" outputs. Another form of peripheral is the display. In a display, a CRT presents imaged information-"soft copy" in computer parlance. Some displays merely present data; others act like teletypewriters, providing alphanumeric images and accepting data from a keyboard input. Others permit the user to act directly upon the image being presented. Quite frequently, minicomputers are required to gather data from a device or sensor that outputs an analog signal or must sometimes send analog signals to a process or device. Converters are used to interface a computer to such a process or device. As its name implies, an analog-to-digital (A/D) converter transforms analog signals into digital form for computer manipulation; digital-to-analog (D/A) converters do the reverse. The key to all peripheral operation is information transfer. The storage units transfer internal to and from the computer. The teletypewriter or its equivalent (often called the input /output or I/O device) transfers instructions or finished data between user and computer. Converters transfer information between the computer and the "real world"--the outside activities the computer is involved in, other than direct communication with the user. When we speak of a computer and a number of peripherals configured together to do a particular job, the result is usually referred to as a "computer system" to differentiate it from its central component. Since the teletypewriter is a peripheral device, a purist often refers to a computer that only has a teletypewriter connected to it as a computer system, although many people consider this particular combination a computer, saving the "system" designation for something a little more elaborate. This has led to some confusion. In this discussion, the author takes the latter approach. An engineer can view a minicomputer in two ways: as a "free-standing" unit to help him with his calculations, designs, and so forth, or as a component to be used in a system to do a particular job. In this latter category, the engineer can consider the minicomputer "component" as something to help in engineering work (such as testing systems), or as something to solve a particular engineering problem (directing a machine to manufacture a consumer product, for example). Although using a minicomputer as a component to solve a particular problem is important to the minicomputer industry, it is only marginally important to the engineer; consequently, emphasis here will be on how the engineer can be helped by a minicomputer-either by itself or in a system--rather than the different ways it can be incorporated. Programming Most engineers who have only a nodding acquaintance with computers consider them simply and solely as devices on which to perform calculations--"supercalculators," in other words. In order to comment upon the "calculating" capabilities of a minicomputer it will be necessary to touch briefly upon the different types of computer programs. "Programming," to many people, means writing certain formulas and statements to enable a computer to solve explicit problems. Actually, it is a little more complex than that. A computer cannot function without some sort of program, but there are all types. The circuits within a computer can recognize an "on-off" relationship (that is, whether a circuit is active or not); "on" and "off" are assigned the numbers 1 and 0 respectively, and thus computer operations are binary, forming the basis of "machine language." A less cumbersome form of programming is an "assembly language" or "assembler." Instructions are read into the computer which enables it to recognize symbolic notation (for example, in a PDP-8 family computer, if you wanted to clear the accumulator, you could type "CLA"; this would be translated by the instructions-program already in the computer-to machine-language binary number 111110000000). Assembler programs are organized the same way machine-language programs are-the way the computer's components operate rather than the way people normally handle mathematics. A form of programming more nearly paralleling the way human language and mathematics operate is known as a "compiler program" or a "compiler." Fortran, to take a well-known example, is a compiler. Because of the organization of a compiler, it is far easier to learn to work with. But you don't get something for nothing. The necessary instructions to convert the compiler's statements into machine language take up considerably more of the computer's memory than an assembler, leaving less of the memory available for the programmer's instructions. Another high-level program is an "interpreter." Such a program is almost English-like in nature and very easy to learn. But, like the compiler, an interpreter takes up quite a bit of the computer's memory. Viewed from the programming context, the minicomputer is a versatile calculating tool. A large number of mini-computers can handle compilers like Fortran or interpreters like Basic or Focal. When a high-level program has been entered in the machine, it is possible to formulate programs in the language that will permit detailed calculations to be performed by an assistant, freeing an engineer for other tasks. In this mode, the program is written by the engineer in such a way that it asks questions of the person. The computer does most of the talking so a relatively unskilled person can work with the computer (i.e., if the computer types out "Please enter the resistance of the load in ohms," the operator only has to enter a figure and, from the question, he knows which figure to enter). In addition to printing out single calculations and programs involving single calculations, minicomputers can perform matrix calculations and solve simultaneous equations (precisely how many depends on the computer in question, how much memory is being used, and what type of program is involved; a 4K PDP-8 minicomputer can handle a 6 X 6 matrix or solve ten simultaneous equations with 10 terms using Focal, for example). Output Devices Minicomputers can be coupled to graphical output devices, permitting them to plot representations of solutions. These can be the form of conventional Cartesian or polar coordinates, and-depending on the computer and program used-can be two- or three-dimensional representations (some minicomputers are capable of generating real three-dimensional plots by plotting a stereoscopic pair of images). A further extension, which actually goes beyond the realm of mere "calculation" for the more complex minicomputer systems, is the use of a visual display and a "light pen" for design work. The "light pen" is a probe with a photosensor built into its tip. The operator holds the pen directly against the face of the CRT; the pen detects the glow of an image on the CRT and sends the information back to the computer via a connecting lead. Since the computer is generating the image, it "knows" what image element of the total scene is being presented. This lets a designer manipulate elements within a design (such as parts layout) and permits the computer to optimize a design (such as circuit routing). Many "display-" configured minicomputers can also be connected to plotters to permit a finished "hardcopy" drawing of the design. As in the case of a high-level language, interactive graphical design is fairly easy to work with. Most systems are designed to be used by someone who is not a sophisticated programmer since the instructions flashed upon the screen are in English. The key to the calculation and design uses of minicomputers is interaction. Many of the programs are set up to permit the user to interact with his minicomputer for maximum effectiveness. Unlike "after the fact" operations such as data-center batch processing (which a minicomputer can also do), the operator can alter his program (or design) while he is working, thus permitting him to operate more creatively. In addition to its calculating abilities, the minicomputer is useful as an engineering instrument for measurement, data logging, and evaluation. Historically, a large number of minicomputers have either been built into measuring systems or interfaced with instruments to extend the scope, precision, or both of such instruments. Engineering Applications Evaluating the performance of engineering prototypes of mechanical and electronic equipment is a normal minicomputer function. The reason a minicomputer is used for this purpose is simple: its operations can be altered by reprogramming; to do the same with a specially constructed instrument would generally require a redesign plus the rebuilding that would be necessary. Thus, while initial costs are similar, the readjustment or modification costs of a minicomputer-based system are lower (with the added advantage that if you wish to return to your original application, you have merely to read in the initial program). In this respect, minicomputers offer an additional dividend. They are general-purpose machines, which means they are able to perform a wide variety of different functions merely by reading in different programs, although the majority of minicomputers. are able to do only one of them at a time. Thus, a minicomputer can be put to several different uses by the same facility. For instance, at one facility, the minicomputer could be used to perform engineering calculations in the morning, system calibration in the early afternoon, and run accounting, payroll, or inventory calculations at night. All it would require is the proper program (and in a few cases, some additional peripheral devices). How Do Minicomputers Compare? Quite frequently individuals who are associated with minicomputers being used for calculations are asked how the machines compare with desk calculators and with large computers used on a time-sharing basis. In one sense, it can be said that a minicomputer has the basic advantages of both with few of the disadvantages of either. In its basic configuration, that is, 4K of core memory and a teletypewriter, the average minicomputer is roughly comparable to a programmable desk calculator with regards to speed and accuracy. It can be used with slightly greater ease, but it is somewhat more expensive. However, the minicomputer has something that the programmable desk calculator has not: expansion potential. The range and type of peripherals available for a minicomputer, plus the memory that can be added, permit a basic minicomputer to be expanded into a reasonably sophisticated data-processing system (and, in addition, it can be used for specialized analytical operations with instruments). By contrast, the expansion of a programmable desk calculator is quite limited. When comparing the minicomputer to time-sharing service, another factor must be considered-the economics of the situation. To be sure, the large time-sharing computer has considerably more power than a minicomputer; however, it is rarely that an engineer needs the very powerful "number crunching" that a time-sharing computer affords. Most routine calculations can be performed on a smaller computer. For that reason, the smaller minicomputer systems--when operating with a high-level language like Focal or Basic--can almost be thought of as the equivalent of time-sharing terminals without the large computer. Since the average rental of a time-sharing terminal will cover the cost of a basic minicomputer system within six months to a year, it usually makes more sense for the user to own his minicomputer. Using individual minicomputers rather than a time-sharing network to perform equivalent calculations offers another dividend besides economy. In a time-sharing network, if there should be a computer malfunction, all terminals would go "off the air." By contrast, independent minicomputers do not face that problem; if one should malfunction, it would not affect the activities of the others. It is the equivalent of not putting all one's eggs in a single basket. Of course, time sharing's chief advantage is that many users have access to the same data base. Indeed, some minicomputer systems are capable of time sharing-and some users employ minicomputer systems for time sharing routinely-but here the large computer system with its greater capacity has a decided edge. But what about those jobs that a minicomputer is unable to handle by itself? A minicomputer can often "go into partnership" with a large computer to solve such problems. In what is known as a "hierarchical" arrangement, a minicomputer can be connected to a large computer over communications lines just like a time-sharing terminal. But the difference in operation is this: a minicomputer will handle all the problems it can cope with by itself. When it runs into the occasional problem beyond its capabilities, it will "pre process" the problem, translating it into machine-language instructions which are, in turn, sent to the large computer. The large computer solves the problem and then sends the answer to the minicomputer, which accepts the problem solution as if it had solved it by itself. The large computer is thus used only when needed and the operating costs of using it can be cut drastically. Minicomputer Costs This brings up the question of over all minicomputer costs. "How much ?" is a question with no easy answer, be cause it can range from very inexpensive to prohibitive. The hardware costs of a minimum configuration are around $6000-$7000 including the Teletype or equivalent input /output device; this will usually include a program that will permit you to get "on the air" with some calculations, although the power and accuracy of the program depends on the manufacturer. The reason costs are hard to pin down, however, is that different people require different configurations. Some just need a desk-sized minicomputer for simple calculations. Some want to store programs or data and re quire a system with a mass-storage de vice like tapes or a magnetic disk. Some need interfacing to instruments. Others may want a visual display. The best rule-of-thumb for deter mining computer-system costs is that, in general, a system will cost about six times the cost of its minicomputer (that is, an average system containing a $5000 basic minicomputer will cost around $30,000); but such a system will be a lot more powerful than a mere calculator. Of these costs, part will be for the additional hardware, part for specialized programming, and part for interfacing devices. The percentages tend to vary with the applications. In some cases, small systems are "expanded" at a later date by adding more equipment to the basic system after it has been in operation for a while. Thus a system can "grow" as extra features are required. Computers have come a long way from the old punch-card monsters living in isolated glory in an air-conditioned comp center. Minicomputers and systems built around them are helping engineers solve their problems "on location." Rather than having to take problems to the computer, the minicomputer is brought to where the problem is. And that can make a big difference in the way a problem can be tackled. |
PREV: | (none) | NEXT | Guide Index | HOME |