168 Part 2 The instruction-set processor: main-line computers
Section 1 Processors with one address per instruction
tenance time. The nonscheduled engineering time has varied from as little as 10.1 hours or 6 per cent to 42.3 hours or 25 per cent. The last column in the Table shows the amount of nonscheduled engineering time as compared to the allowable operating time (total time less preventive maintenance time). Here there is a variation of from 7.6 to 31.7 per cent and an average for the weeks shown of 16.9 per cent. It is believed that these figures, while good for the first months of operation of a new piece of equipment, will show definite improvement over the next year.
Although the opportunity to prove or disprove the following theory of operation has not presented itself, it is believed logical that optimum use of the UNIVAC equipment might be obtained by means of scheduling preventive maintenance only at such times as it is indicated in the judgment of competent operators. In other words, there are many occasions preceding a scheduled maintenance shift when the system is performing very well. At such times, it is extremely inefficient to shut down the operation in order to provide maintenance. For many reasons, however, it has been impossible to operate and maintain the first system in this way. It is hoped that such operation will be possible in following installations.
It should be realized that the UNIVAC system requires a supervisor of the same caliber as the one required for a large punched card installation. However, the large group of operating personnel would be replaced by a small group of well-trained extremely competent people thoroughly familiar with the details of the computer and associated equipment. The time spent in providing a high degree of training for these people is more than repaid in increased operating efficiency and consequently higher work output. For example, situations arise in the course of running a problem where a correct operational decision can save hours of elapsed computation. Also, a competent operator will recognize malfunctions sufficiently early to prevent serious delays. He is capable of deciding whether to continue with machine operation or to stop to diagnose. The second UNIVAC system which is ready for installation in Washington, will be operated by a group of engineers who have been trained in operation and maintenance. This procedure, it is believed, will result in the UNIVAC system being of maximum benefit to the Air Comptroller's Office.
Evaluation of UNIVAC design
Checking features
Maintenance of the UNIVAC has been vastly simplified by use of duplicate arithmetic and control equipment and other checking methods. Many factors which would have led to undetected errors have, by virtue of duplication, immediately stopped the computer. Although checking by means of inverse operations can provide operational checks on the arithmetic circuits there is some question as to whether it provides as good a check as duplication. However, in connection with odd-even codes, it may conceivably be comparable. It should be remembered, however, that this is from an operational standpoint and not a maintenance standpoint. When the control equipment is considered it is difficult to visualize a check that is as good as duplicated equipment. Other checks that are utilized in UNIVAC include the periodic memory check, intermediate line function table checker, function table output checker, memory switch checker, and 720 checker.
As explained earlier in the paper, the periodic memory check is accomplished by reading out of all memory channels sequentially and performing an odd-even check on each digit as it passes through the high speed bus amplifier. The period at which the check is repeated may be varied over a large interval. At present, it is set at 5 seconds, the check taking 52 milliseconds or about 1 per cent of the computing time.
The function table has a check at the very input by bringing in the check pulse in each character so that if an odd-even error occurs between the control register and the static register, no order will be set up and the computer will grind to a halt! If the input sets up properly but an error occurs farther on in the table, but not ahead of the intermediate lines (the linear set into which the input combinations are decoded), the error is caught at this point. The intermediate lines are broken into groups in such a way that an error is indicated when more than one line is set up in one group or the entire set. There is an exception to this in some groups where no error is indicated by this checker if more than one line is set up within the group.
This has been allowed only in those cases where it has been shown that setting up two or more lines will cause some other checker or checkers to indicate the trouble.
If the error occurs beyond the intermediate lines, the output checker then comes into play. This checker makes an odd-even count on the number of gates used on each instruction: dummy lines having been added so that the count is normally always odd.
The memory switch or tank selector checker ensures that one and only one memory channel is selected on any instruction. It checks each of the two digit positions separately indicating which if either, is in error.
The 720 checker counts the digits coming off the tape and if there are either more or less than 720 in one block, the computer stops; by examining the indicators on the supervisory control console, the operator can determine the number of digits actually